Potato is a cool season crop susceptible to higher temperatures. Chlorophyll fluorescence efficiency measurement of photosystem II (PSII) is an effective and non-invasive technique to detect damage due to heat stress. Thus, the effects of long exposure to two range of temperatures in the dark at control (14-24oC) and heat stress (24-34oC) were investigated in 21 Embrapa Potato genebank accessions by means of chlorophyll fluorescence, quantum yield of PSll (Yll), non-photochemical fluorescence quenching (NPQ), coefficient of non-photochemical fluorescence quenching (qN), coefficient of photochemical fluorescence quenching (qP), electron transport rate (ETR), photosynthesis rate (Pn), stomata conductance (Gs), transpiration rate (Tr) and maximum quantum efficiency of PSII (fv/fm). Genotypes under stress presented distinct levels of inhibition of the photosynthesis rate. Based on fv/fm, the genotypic values allowed rank the first and last performing genotypes of S. commersonii, S. chacoense, and dihaploids of S. tuberosum. Selection was based on the Best Linear Unbiased Prediction (BLUP) model. Residual maximum likelihood (REML) measured component of variance suggested that the estimated traits were worth to investigate and gives 68% accuracy in selection of genotypes. Analysis of variance shows significance among genotypes except for Gs. Pearsonâ€™s correlation of fv/fm was also checked for both temperature ranges. Under control, none of the measured traits show significant correlation with fv/fm while under stress fv/fm was positively correlated with Y(ll), qP and ETR and moderate negatively correlated with NPQ and qN. S. chacoense accessions showed higher Pn, Gs and Tr, followed by S. commersonii and S. tuberosum dihaploid. Gs of wild genotypes were lower under stress while Pn were found higher. Interestingly, S. chacoense accessions showed 75% lower transpiration rates when compared to accessions of the remaining species. Thus, we confirm that Fv/Fm is an effective selection tool for heat-tolerant and heat-sensitive wild genotypes, based on the mixed model analyses.