In this paper a procedure for designing structures under uncertainty is presented. The uncertainty model of fuzzy randomness is employed to take into account the uncertainty of structural parameters in a realistic and comprehensive manner. This uncertainty model includes real valued random variables and fuzzy variables as special cases. Objective uncertainty and subjective uncertainty are processed simultaneously. Algorithms of fuzzy structural analysis (processing of fuzzy variables in structural analysis) and fuzzy probabilistic safety assessment (processing of fuzzy random variables, real random variables, and fuzzy variables in safety assessment) are used to compute fuzzy structural responses and fuzzy safety prognoses, which are the backbone of the new design concept. Comparing fuzzy structural responses and the fuzzy safety level with permissible values, discrete permissible and nonpermissible parameter vectors are identified. These are introduced into a fuzzy cluster analysis to obtain permissible and nonpermissible clusters (continuous sets of real parameter vectors with similar properties), which represent the basis for generating uncertain structural design parameters. This concept is referred to as fuzzy cluster design. It can be combined with arbitrary fundamental solutions for deterministic and probabilistic structural safety analyses. For instance, well developed algorithms of Monte Carlo simulation and codes of nonlinear structural analysis can be incorporated in the procedure. The algorithm of fuzzy cluster design is presented in detail and demonstrated by way of a numerical example.