The performance of banana and plantain hybrids and cultivars must be assessed, across locations and over several crop cycles. The objective of this research was to define the optimum plot configuration for . Musa field-testing based on the repeatability of yield and phenological characteristics across and within locations in West Africa. Datasets of . Musa multi-environment trials available from the International Institute of Tropical Agriculture were used for this study. Plant height and bunch weight showed the highest repeatability at two Nigerian locations, while days to harvest and fruit filling time had on average the lowest repeatability across locations. Plant height was the characteristic requiring the smallest plot among all traits, irrespective of location. This article provides a table showing the relationship between sample size (. N), repeatability (. R) and level of confidence for phenotypic discrimination of . Musa breeding materials. For example, if one wishes to assess phenotypic values with an accuracy of 95% and the correlation between two measurements of phenotype is 50% (. R=. 0.50) then the sample size required is . N=. 10. © 2012 Elsevier B.V.