To investigate the poor participation rate of cranberry growers in the multiple-peril crop insurance program, a sample of 15 Massachusetts growers was interviewed. According to their risk preferences, a much greater proportion of growers should have insured, than actually did. A possible solution is to match the distribution used by the insurer closer to that believed by the grower. Adjusting each grower's historical yield series for trend brought the historical and subjective mean yields much closer. However, an aggregate test found the effect of adjustment to be insignificant, implying that the avenue for increased participation lies elsewhere.