A Song of Ice and Fire Roleplaying

A Song of Ice and Fire Roleplaying (in following: SIFRP) is the second official pen-and-paper adaptation of George R.R. Martin's unfinished low fantasy epic, following A Game of Thrones (2005) by the defunct Guardians of Order. Designed by Robert J. Schwalb for Green Ronin Publishing, the game was originally released in 2009, then revised and updated with the A Game of Thrones Edition in 2012, capitalizing on the success of the televised adaptation of the novels, Game of Thrones (2011-2019).

A note on the translation: Unlike with most games in this survey, the only version of the SIFRP rulebook available to the author at the time of writing was the 2013 German translation of A Game of Thrones Edition by Mantikore Publishing, which will be referenced henceforth. This also accounts for any term inconsistencies between the original English-language rules and the text below.

Free resources: Quick-Start.

Core Mechanic

The core mechanic of SIFRP (referred as an "ability test", pp. 9-24) uses pools of d6s to resolve player actions. The simplified procedure is as follows:

  1. The player declares their character's objective and how they attempt to achieve it.
  2. The GM and the player negotiate which primary ("ability", ranging from 1 to 7) and secondary stat ("specialization", ranging from 0 to 7) applies to the attempted action.
  3. The GM sets the "difficulty" of the test, ranging from 3 (Easy) to 21 (Heroic).
  4. The player rolls a number of d6s equal to their character's relevant ability score, plus any "bonus dice" granted by their secondary stat and circumstantial factors.
  5. If any bonus dice have been rolled, the player now discards the lowest results from the roll until only a number of dice equal to the PC's relevant ability score remains.
  6. The player sums up the remaining results and adds or subtracts any further numeric "modifiers" granted by their character's traits and circumstantial factors.
  7. If the final result from step 6 equals or exceeds the difficulty from step 3, the PC achieves their objective (Yes); otherwise, their action fails (No). The gradient of success is based on how much the test result exceeds the difficulty, ranging from 1 (if result exceeds difficulty by 0-4) to 4 (15+) and representing a finer gradation of Yes-And outcomes. This "degree of success" has many mechanical effects in various game subsystems.
  8. The GM describes the outcome of the character's action.

In addition to the basic ability test, the game features three specialized variations. An "extended test" requires the player to succeed at several basic tests in a row, with each attempt taking some span of in-fiction time, before their character can reach their declared objective. A "competition test" is used when two characters compete to reach the same objective: in this case, they roll with the same stats against the same difficulty, and the character with the higher degree of success wins. Finally, the "conflict test" is used for any action that can be considered an "attack" and differs from the basic test mainly in that the difficulty in step 3 is derived from the "defender's" stats, instead of being set by the GM.


Given its source material's heavy emphasis on political intrigue and conspiracy, it isn't surprising that SIFRP provides the most elaborate ludic procedure for resolving social interactions out of all titles in this survey. Social conflict (referred as "intrigue") is simulated in much the same manner and detail as physical battle: an intrigue mechanically consists of one or more "exchanges" (equivalent to "rounds" in turn-based combat, although their in-fiction duration is highly variable, whereas combat rounds typically represent in-fiction time spans of a fixed length), during which each active participant, PC or NPC, gets a single turn in which they can perform a limited number of actions (again, equivalent to turn-based combat).

In physical combat, as simulated by SIFRP, an attacker makes conflict tests with their Fighting and Marksmanship abilities against the target's "combat defense" score in order to deal "damage" and to reduce their "health" (hit points); in an intrigue, the agent makes conflict tests with their Persuasion and Deception abilities against the patient's "Intrigue Defense" in order to "influence" the patient and to reduce their "Composure". The Intrigue Defense is the sum of the patient's Awareness, Cunning, and Status ability scores, whereas their starting Composure equals their Will ability score times 3 and is reset to the maximum at the start of every intrigue.

The main primary stats (abilities) used by the social agent in an intrigue are "Persuasion" and "Deception", with the latter used mainly when deceit as their current objective or when the agent employs a technique normally used with Persuasion as a bluff rather than in earnest. Both abilities have a number of subordinate secondary stats (specializations): "Bargain", "Charm", "Convince", "Incite", "Intimidate", "Seduce", and "Taunt" under Persuasion and "Act" and "Bluff" under Deception.

The ten-step intrigue procedure is described on pp. 148-161 (which is reduced to just five steps in the free Quick-Start guide, pp. 17-21):

  1. Type. The GM adjudicates the scope of the intrigue:
    1. "Simple intrigues" play out between the agent and the patient in a single turn, with the agent's ability test result determining the short-term outcome. This procedure is roughly equivalent to simple Charisma checks in D&D.
    2. "Regular intrigues" play out between two or more participants over multiple exchanges until all but one side either "yield" (see below) or are "defeated" by having their Composure reduced to 0. The winner(s) then achieve their declared objective (see step 3).
    3. "Complex intrigues" play out between multiple participants pursuing far-reaching objectives and consist of multiple regular intrigues, with "victory points" scored by winning them and lost by being defeated. The first side to reach the victory point threshold set by the GM (ranging from 3 to 6 or more) achieves its objective, while the others fail.
  2. Scene. The GM describes the location where the intrigue takes place (which can grant situational bonuses to individual participants' Intrigue Defense) and who is participating in it. Additional participants may join the intrigue later on, but doing so effectively resets it, negating any progress made by everyone until then.
  3. Objective. The player of each character participating in the intrigue (the GM for NPCs) declares the immediate objective their character pursues in it. This step is notably absent from the equivalent combat procedure, since the attacker's objective is always, implicitly, to achieve physical dominance over the defender, and the actual outcome for the loser is decided by the winner after their victory. Intrigue objectives can be changed at the start of every exchange, but doing so restores some of the patient's Composure. Typical objectives outlined by the rules include:
    1. Friendship: The agent wants to establish or to improve a beneficial relationship with the patient.
    2. Service: The agent wants the patient to carry out a specific task for them.
    3. Information: The agent wants to learn a specific piece of information the patient is privy to.
    4. Deception: The agent wants the patient to believe a specific lie or fabrication. Choosing this objective requires the agent to use their Deception ability against the patient (others rely on Persuasion).
  4. Disposition. Each participant's player quantifies their current attitude towards every other participant by assigning to it a "Disposition Rating" (DR) – ranging from 1 ("affectionate"), through 4 ("indifferent"), to 7 ("malicious"), – according to a number of factors, such as personal attractiveness (erotic capital), reputation, past interactions, history, and currently declared objectives.
    Beyond setting the basic expectations for the tone of role-played conversations, DR two ludic effects: whenever the agent influences the patient, the actual amount of Composure they lose is reduced by their current DR towards the agent, potentially to 0 (this is equivalent to the "damage reduction" by armor in combat); additionally, lower DRs give bonuses to Persuasion and penalties to Deception tests, and vice versa.
    After the first  exchange between two characters, their respective players may shift their DRs towards each other up or down by 1 step at the start of each subsequent exchange between them. However, if one of them has successfully influenced the other, the patient's DR towards the agent may not worsen at the start of their next exchange (but it may improve).
  5. Initiative. Each participant makes a Status ability test, and the one with the highest result takes the first turn in each subsequent exchange, then the next highest, and so on. Like in combat, the player may also "delay" taking their turn until after a lower-initiative participant has taken theirs.
  6. Technique. On their turn, the agent player selects how they attempt to influence the patient, with the chosen "technique" determining which of the agent's stats governs the extent to which they can influence the patient with a successful ability test, as well as which specialization applies if they make a Persuasion or a corresponding Deception test. The seven available techniques are roughly equivalent to the choice of weapons in combat and correspond to the seven Persuasion specializations:
    1. Bargain: The agent tries to negotiate a profitable exchange of goods or services with the patient.
    2. Charm: The agent tries to improve the patient's DR towards them by 1 step.
    3. Convince: The agent tries to gain the patient's support with rational arguments.
    4. Incite: The agent tries to temporarily worsen the patient's DR towards someone else.
    5. Intimidate: The agent tries to force the patient to retreat or else to set their DR towards the agent to 3, ensuring their full compliance as long the two of them remain in proximity.
    6. Seduce: The agent tries to temporarily improve the patient's DR towards them, potentially initiating an intimate liaison.
    7. Taunt: The agent tries to goad the patient into performing a specific action, at the cost of worsening the patient's DR towards them.
  7. Role-playing. On the agent's turn, their and the patient players role-play the actual dialogue between their characters, with the GM rewarding convincing performances with bonuses and penalizing obvious blunders (but not poor role-playing if the players are obviously uncomfortable with performing in-character).
  8. Actions and Tests. On their turn, the agent player may perform a single "action", equivalent to various actions in turn-based combat, making the corresponding ability test for them. The available actions are:
    1. Influence: The agent tries to influence the patient and to reduce their Composure. This is equivalent to the basic "attack" action in combat, with a successful conflict test against the patient's Intrigue Defense potentially reducing the latter's Composure (it will be discussed in detail below).
    2. Assist: The player makes a basic Persuasion test that, if successful, gives a bonus to the next conflict test made by an ally of their choice.
    3. Consider: The player skips their action to gain 2 bonus dice for their action on their next turn.
    4. Fast talk: The agent tries to confound the patient with a stream of nonsense, with a successful conflict test temporarily reducing the patient's Intrigue Defense.
    5. Manipulate: The agent tries to goad the patient into using a technique of the agent's choosing on the patient's next turn.
    6. Mollify: The player makes a basic Persuasion test that, if successful, restores some lost Composure to an ally of their choice.
    7. Quit: The player may have their character abandon the ongoing intrigue, as long as they have a venue for escape. Doing so, however, often comes with strings attached, at the GM's discretion.
    8. Read target: Once per intrigue, the player may make an Awareness conflict test against an opponent's Deception to gain insight into their DR and technique. If successful, they gain a bonus die for all conflict tests against that opponent until the end of the intrigue.
    9. Shield of Reputation: Once per intrigue, the player may make a Status, which, if successful, improves the patient's DR towards the PC by 1 step.
    10. Switch to Combat: The player abandons the ongoing intrigue by escalating it to physical combat. There is no "Switch to Intrigue" action in the combat rules, precluding a de-escalation of conflict.
    11. Withdraw: The player makes a basic Will test to temporarily improve their Intrigue Defense.
    12. Yield is a special action that can be attempted in addition to the above. See below for the specifics.
  9. Repeat. Unless one party has emerged as the clear winner, the regular/complex intrigue continues into the next exchange, with the procedure returning to step 2 (i.e. potentially changing the scene).
  10. Outcome. Once one of the parties has achieved a clear victory, the intrigue is adjourned and the specific outcome is adjudicated and narrated by the GM.

Whenever an agent employs the "Influence" action in step 8, they make a conflict test with Perception or Deception and a specialization determined by the technique they chose in step 6 against the patient's Intrigue Defense score. If successful, they deal an number of "influence points" (our term) equal to the degree of their success times the technique-dependent ability score (Awareness, Cunning, Persuasion, or Will). These influence points are then reduced by the patient's DR towards the agent (but not below 0), and the remainder is deduced from the patient's Composure.

The patient player may choose to reduce the influence points inflicted upon them even further by taking points of "Frustration". Each Frustration point reduces the Composure loss by an amount equal to the patient's Will ability score, but also cumulatively removes one dice from all subsequent Persuasion and Deception checks. Frustration points are reset to 0 at the end of the intrigue, regardless of its outcome. This mechanic is equivalent to Injuries and Wounds in the combat rules (pp. 177-178), although the latter are much more lasting and require lengthy recovery.

If the patient's Composure is reduced to 0 or their Frustration points exceed their Will ability score, they are "defeated" (i.e. removed from the current intrigue), and the last party standing is subsequently declared winner and achieves their declared objective. On their turn and only once per intrigue, a losing participant may also attempt to "yield" to the opposition by offering them a compromise outcome, which the opposition may accept or come up with a counter-offer. If either of the offers is accepted, the yielding participant immediately leaves the intrigue; otherwise it proceeds as normal. The same option is also available under the combat rules (p. 176).

We observe that the wording of the procedure heavily implies that player characters work together against GM-controlled NPCs, alternating between agent and patient roles with each exchange (except in simple intrigues, which appear to be reserved for the "PC agent, NPC patient" configuration). The same rules, however, are also perfectly applicable to the "PC agent, PC patient" configuration, with players intriguing against each other, just as they would engage each other in an (in-fiction) battle. Player agency is effectively preserved by treating persuasive interactions as a form of highly formalized turn-based combat, while simultaneously offering players mechanical options to abort an intrigue by quitting, escalating it to actual combat, or by yielding to negotiate a compromise.