If genotype (G) ranks change from one environment (E) to another, genotype by environment interactions (GEI) reflects the need for testing Gs in numerous Es in order to obtain reliable results. The aim of this study was to compare 16 non-parametric stability statistics (NPSSs) for GEI on grain yields of 15 durum wheat genotypes, consisting of 11 advanced lines selected from Turkish National Durum Wheat Breeding Program (TNDWBP) and four checks, tested in 12 rain-fed environments during the 2 cropping seasons (2009–2010 and 2010-2011) in Turkey. The combined ANOVA indicated that G, E and GEI effects were significant for grain yield. According to analyses of NPSSs, the highest in ranking (TOP), percentage of adaptability (PA), rank mean (RM) and yield-stability (YS) statistics were positively associated with grain yield and therefore characterized under the dynamic concept of stability. In addition, spearman rank correlation analysis revealed that only TOP, PA, RM and YS statistics would be useful for simultaneous selection for high grain yield and stability. Based on the 16 NPSSs used in this study, Dumlupinar cultivar (G15) was both the most stable and one of the highest yielding ones. On the other hand, G5 and G7 were the most stable ones among the advanced lines tested, but their yield performances were lower. As a result, this study showed that the crossing block of TNDWBP should be enriched by germplasm being capable of dynamic stability, wide adaptation and higher yielding.