Current models of biological communication point at evolutionary mechanisms of particular signal types. Those that present complete models look at the signals' equilibrium values and their evolutionary stability, and require two simultaneous equations: an equation that describes the signaler's fitness as a function of the signal and of the recipients' response, and a simultaneous equation that represents the fitness of recipients. This paper examines the effect of different signal types, such as handicaps, amplifiers, camouflage, mimicry etc, on the first equation. By considering parameters that affect the evolution of signals this paper first constructs a general model of biological signaling. Different signal types are then characterized by different sets of limiting assumptions. As a result, the fitness of a signaler of each signal type is represented by a unique equation that is a mathematical derivation of the general signaling model. This analysis enables a natural division of signals into groups and subgroups that share similar assumptions and properties. It shows the importance of signal design, and points at three methods by which signals may be reliable: by trade-offs between cost and benefits, by design and by convention. Copyright 1997 Academic Press Limited