One of the most common shortcomings in non-communicable disease risk factor surveillance, especially in prevalence studies, is sampling procedure, which can and does compromise accuracy and reliability of derived stimates. Moreover, sampling consumes significant time and resources. Since the early 1990s, risk factor surveys in Cienfuegos province, Cuba have paid particular attention to careful sampling methods. The new survey conducted in 2011 was not only statistically rigorous but introduced an innovative, more effi cient method. This article provides a detailed description of the sample design employed to optimize resource use without compromising selection rigor.