Plant populations and species differ greatly in phenotypic plasticity. This could be because plasticity is advantageous under some conditions and disadvantageous or not advantageous under others. We distinguish adaptive from injurious and neutral plasticity and discuss when selection should favor adaptive plasticity over genetic differentiation or lack of phenotypic variation. It seems reasonable to hypothesize that selection is likely to favor plasticity when an environmental factor varies on the same spatial scale as the plant response unit, when the plant can respond to an environmental factor faster than the level of the factor changes, and when environmental variation is highly but not completely predictable. Phenotypic plasticity might also tend to be more advantageous when mean resource availability is high rather than low, when a response can occur late in development rather than early, and when a response is reversible rather than irreversible. There is substantial evidence for the hypothesis that predictability favors plasticity. However, available evidence does not support the hypothesis that high mean resource availability necessarily favors plasticity. Testing hypotheses about when it is good for a plant to adjust is central to understanding the diversity of plasticity in plants.