Mining of new genetic resources is of paramount importance to combat the alarming spread of stripe rust disease and breakdown of major resistance genes in wheat. We conducted a genome wide association study on 352 un-utilized Afghan wheat landraces against stripe rust resistance in eight locations. High level of disease variation was observed among locations and a core-set of germplasm showed consistence performance. Linkage disequilibrium (LD) decayed rapidly (R2≈0.16 at 0cM) due to germplasm peerless diversity. The mixed linear model resulted in ten marker-trait associations (MTAs) across all environments representing five QTL. The extensively short LD blocks required us to repeat the analysis with less diverse subset of 220 landraces in which R2 decayed below 0.2 at 0.3cM. The subset GWAS resulted in 36 MTAs clustered in nine QTL. The subset analysis validated three QTL previously detected in the full list analysis. Overall, the study revealed that stripe rust epidemics in the geographical origin of this germplasm through time have permitted for selecting novel resistance loci.