School improvement is much sought and often claimed. However, it is questionable whether overall achievement in countries such as the USA or England has improved by any significant amount over thirty years. Several school improvement programmes have been claimed as successful, but evaluations, even where they exist, are generally poor: based on the perceptions of participants, lacking any counterfactual or reporting selectively. Accounts of improvement in individual schools are numerous, but are inevitably selective; the attribution of causality is problematic and knowledge of the conditions under which such phenomena are likely to be replicated is limited. School effectiveness research also has yet to identify specific strategies with clear causal effects. In short, many claims of school improvement are illusory. Nevertheless, there are some improvement strategies that are well-defined, feasible and robustly shown to be effective. In future, we need greater clarity and agreement about what constitutes success. Evaluation must be taken more seriously, and its results treated more critically.