Interspecific mutualism often involves partners trading different kinds of services. The cost and benefits of the trade may differ from one interaction to the next, or fluctuate over time in a single interaction, due to factors both internal and external to the mutualism itself. Aspects of partner quality, such as a partner's ability to provide services and its need for the services obtained in return, are likely to be particularly important. As a consequence of such variation, one might expect strategies of adjusting investments to the characteristics of the partner and to other circumstances, in a way that increases the benefit received and/or decreases the investment necessary to obtain the benefit. Lycaenid butterfly larvae often interact mutualistically with ants, which they reward with nutritious secretions in return for being protected from parasitoids and predators. The larvae also have other means of influencing ants, such as a pair of eversible tentacular organs, which emit volatile substances that attract and alert ants. Both participants in the interaction can readily modify their mutualistic investments over very short periods of time. My aim in this thesis has been to investigate strategies individual larvae could use to change the cost-benefit relationship of the interaction in a way that increases their returns. For this purpose I have manipulated factors that might influence the outcome of the mutualism for lycaenids and observed their response, both as tentacle signalling and in the form of investment in secretions. Examples of factors I have investigated are the level of ant attendance, the need for protection, and the species of the attending ants. I have also investigated the effect of group size and companion quality in a gregarious lycaenid species. I have found that larvae react to such changed circumstances by varying their behaviour, either acting to increase the number of attending ants when better protection is likely to be needed or reducing the costly secretion when adequate protection is already available. My overall conclusion is that lycaenid larvae have a series of adaptations that serve to efficiently regulate the number of protecting ants and that much of the variability I have observed can be interpreted in this way.