Mutualisms teeter on a knife-edge between conflict and cooperation, or so the conventional wisdom goes. The costs and benefits of mutualism often depend on the abiotic or biotic context in which an interaction occurs, and experimental manipulations can induce shifts in interaction outcomes from mutualism all the way to parasitism. Yet, research suggests that mutualisms rarely turn parasitic in nature. Similarly, despite the potential for ‘cheating’ to undermine mutualism evolution, empirical evidence for fitness conflicts between partners and, thus, selection for cheating in mutualisms is scant. Furthermore, mutualism seldom leads to parasitism at macroevolutionary timescales. Thus, I argue here that mutualisms do not deserve their reputation for ecological and evolutionary instability, and are not on the verge of breakdown.