Reliability-based design of wind turbine blades requires identification of the important failure modes/limit states along with stochastic models for the uncertainties and methods for estimating the reliability. In the present paper it is described how reliability-based design can be applied to wind turbine blades. For wind turbine blades, tests with the basic composite materials and a few full-scale blades are normally performed during the design process. By adopting a reliability-based design approach, information from these tests can be taken into account in a rational way during the design process. In the present paper, a probabilistic framework for design of wind turbine blades are presented and it is demonstrated how information from tests can be taken into account using the Maximum-Likelihood method and Bayesian statistics. In a numerical example, the reliability is estimated for a wind turbine blade in both ultimate and fatigue limit states. Information from tests is used to formulate the stochastic models used in the limit state equations. Partial safety factors for use in traditional deterministic design are estimated using the stochastic models.