We analyse area yield insurance contracts and weather index insurance contracts in terms of risk reduction for wheat farms in Kazakhstan over the period 1980-2002. We combine the common mean-variance approach with a stochastic dominance approach to ensure the consistency of the empirical results with expected utility theory. Results indicate (1) the need for combining both approaches, because for some insurance schemes, the mean-variance results of one-third of the farmers are not necessarily consistent with EU theory. (2) Bootstrapping shows that an expected positive variance reduction is not statistically significant for up to one-third of the farms. Both results indicate that previous methods probably overestimate the effectiveness of crop yield and weather index insurance schemes, in particular for insurance schemes with basis risk. From a practical point of view, (3) area yield insurance based on the county (rayon) yield provides substantially higher variance reduction than reported in the literature, indicating that area yield insurance contracts might be more appropriate in Kazakhstan because of the high systemic yield risk there - an effect of exposure to drought. (4) There are no substantial differences in the results generated by means of different weather indexes. (5) However, compared to farm yield insurance with a low strike yield in order to reduce moral hazard, weather index insurance can be a reasonable alternative for farmers.