Two main aims of this investigation were to predict kernel ratio (KR) and kernel weight (KW) from some walnut characteristics, respectively. For these aims, a total of 112 Walnut genotypes growing in nature were collected at Darende District of Malatya province in the Eastern Anatolia region of Turkiye. The walnut characteristics evaluated were nut length (NL), nut width (NW), nut height (NH), nut weight (NWe), shell thickness (ST), kernel ratio (KR) and kernel weight (KW), respectively. Independent variables were subjected to factor analysis based on principal component extraction method and VARIMAX rotation. On the basis of jointly using factor scores in multiple regression, KR (81.3 % R2 and 80.6 % adjusted R2) and KW (94.7% R2 and 94.5% adjusted R2) characteristics were predicted by using four factor scores with a big accuracy without multicollinearity problem. Consequently, the present results revealed that, walnuts of heavier KW and NWe in the prediction of KR would be expected to produce those of higher KR, and walnuts of higher values in NH, NW, NWe, ST, NL, and KR in the prediction of KW would be expected to produce those of heavier KW. The knowledge may help walnut breeders to improve new selection strategies.