The objective of this study was to test if the response to 10 cycles of a recurrent selection program conducted under conventional tillage and rain fed conditions was the same when contrasted, for several traits, under conventional and non-tillage practices. During two season (2011 and 2012) the 44 S-derived families (four/C 0 to C 10 populations) were evaluated under conventional and non-tillage systems in two fields next to each other. Days to anthesis, plant height, grain, and biomass yield and 1000-grain weight were determined. The grain number per m 2 and harvest index was also estimated. From a random sample of 10 tillers the spikelet per spike and grains per spike were measured. For each trait, a linear mixed model (regression) was fitted to the experimental data. The slopes, under conventional tillage, were significant greater than zero for grain yield, harvest index, seeds per square meter, spikelet per spike and seeds per spike. Under non-tillage the list of traits showing slopes significantly greater than zero was shorter. For most traits there was a significant difference in the intercept terms between conventional tillage and non-tillage, which is interpreted as the tillage-practice effect. The concurrent evaluation in conventional and non-tillage soil managements of ten cycles of a recurrent selection program performed under conventional tillage confirmed the occurrence of a significant genetic progress only under conventional tillage.