Competitive altruism is a possible mechanism for the persistence of cooperative behaviours, specifically those that are performed unconditionally. The theory of reciprocal altruism can be used to explain behaviours that are performed by a donor who receives some sort of benefit in the future. When no such compensation is received, however, reciprocity fails to explain altruistic behaviour. Both reciprocal altruism and competitive altruism anticipate that the more altruistic one is, the more they will receive. Competitive altruism explains unreciprocated altruistic behaviour as individuals striving to outcompete others in terms of their generosity so as to gain the unique benefits obtained from an enhanced status and reputation.

Compared to other primates, humans show a much higher degree of altruism and cooperation towards unrelated individuals. This behaviour is unusual as it goes against one's best interests of benefiting oneself and their relatives in the interest of better survival. A study using sharing games to investigate the ontogenic origins of competitive altruism found evidence that a significant developmental change occurs in children from 5 to 8 years old. The 8 year old participants were more generous in the sharing game, especially when they were observed and their behaviour could affect their chances of being partnered with. The differences in behaviour between the 5 and 8 year olds suggest there is a component of their development at this stage that allows them to learn the mechanism of competitive altruism.

Evolutionary psychologists believe that altruistic behaviour provides adaptive advantages to humans. For example, through self-sacrificial competitive altruism, individuals perform conspicuous self-sacrificial prosocial acts to promote their desirable qualities. This allows them to be viewed favourably by others, which may reap benefits such as a desirable job or better choice of mate. A study found that sex played an important role in triggering this behaviour. Participants were placed into mixed-sex trios to complete a series of tasks, which involved self-sacrificial actions that were viewed as costly, difficult, and crucial towards the group's success. Participants who engaged in self-sacrificial behaviour were favourably viewed, and were rewarded later on by other participants with more money and being preferred as a task partner. The males scored higher than females for the self-rated personality trait of glory seeking, whilst females scored higher for social inhibition. Additionally, the more glory seeking males tended to engage in the self-sacrificial behaviour. Furthermore, the presence of other males seemed to trigger competitive altruistic behaviour, with males oftentimes pushing females out of the self-sacrificial roles, despite the willingness of several female participants. Moreover, males in the self sacrificial roles perceived their role as being of higher status than females in the same role did. This role's desirability can be attributed to it being a conspicuous display of self-sacrificing altruism that exemplifies their helpfulness and ability to withstand the cost. This behaviour can also be highly beneficial for males during mate selection. For example, many bird species' males have elaborate plumage. Although this reduces other aspects of the male's fitness, it is suggested that other than making themselves visually attractive, it signals their superior resistance to parasites. This is because a more elaborate plumage display indicates they can afford to divert those resources to their appearance instead of their health.

To explain competitive altruism, Roberts uses the example of preening among birds. Because certain birds cannot reach parasites on all parts of their bodies, particularly their necks, they benefit from preening one another. For any given bird, there is an entire flock of potential preeners, who compete in hopes of establishing a beneficial relationship. Cheaters, or those birds that try to be preened without preening others, do not compete and thus are excluded from these relationships. Their fitness is lowered because they are ostracized by members of the flock.

McNamara et al. quantitatively analyzed this theory. Like Robert Axelrod, they created a computer program to simulate repeated interactions among individuals. The program involved players with two genetically determined traits, a "cooperative trait" and a "choosiness trait". They found the following results:

'Paradoxical' trait combinations yield particularly low payoffs: individuals with low choosiness but high effort tend to get exploited by their co-players; individuals with high choosiness but low effort waste their time searching for better co-players, which are, however, unlikely to accept them. The positive correlation between choosiness and cooperativeness leads to a positive assortment between cooperative types – an essential feature of all mechanisms that promote cooperation.

The development of such cooperation requires variation in the degree of cooperation and choosiness, which the researchers attributed to genetic mutation and variation. McNamara et al. also determined that since a period of searching is required for "mutually acceptable" players to find one another, competitive altruism is more likely to arise in animals with long life spans.

To relate this condition to the prisoner's dilemma, an individual may benefit the most in a one-time interaction with another by defecting (i.e. receiving benefits without incurring any cost to itself). However, in an iterated prisoner's dilemma, where individuals interact more than once if the act of defecting makes the individual less likely to attract a fit mate in the future, then cooperative behavior will be selected for.