We describe theoretical and experimental studies that demonstrate that a given pattern of neuronal activity can be produced by variable sets of underlying conductances. Experimental work demonstrates that individual identified neurons in different animals may show variations as large as 2-5 fold in the conductance densities of specific ion channels. Theoretical work shows that models with this range of variation in many of their maximal conductances can produce similar activity. Together, these observations suggest that neurons and networks may be less tightly tuned than previously thought. Consequently, we argue that instead of attempting to construct single canonical models of neuronal function, it might be more useful to construct and analyze large families of models that give similar behavior.