A sampling protocol for detecting latent infections in potato tubers stored in bins, based on the preparation of standard samples of 200 tubers, was tested for reliability. Bins of tubers were manually distributed in windrows and 1% of tubers were replaced with painted tubers, simulating tubers with latent infections. The painted tubers were distributed in windrows with clustered, regular and random distribution. Tubers were harvested and transferred to bins. Each bin was considered to be divided into layers 15 cm in depth, and 20 tubers were collected along an X-shaped sampling path of each bin layer. A simulated sampling approach was applied to obtain 10 bulk samples (200 tubers) for each kind of tuber distribution. The number of marked tubers, the mean percentage and the standard deviation of the bulk samples of each group was computed. The results confirmed that the sampling design adopted in this study ensured preparation of representative bulk samples. In fact, marked tubers were effectively detected; the observed mean percentages of marked tubers were acceptable in value, even though an overestimate of the true percentage was generally obtained. The percentage of marked tubers was underestimated in 50% of cases when considering a regular distribution of marked tubers. However, it was never underestimated when considering the clustered distribution, which is most representative of the situation for tubers with latent infections.