A sample of 249 skeletons (154 males, 95 females) from the Chiang Mai University Skeletal Collection was studied to investigate the potential of proximal hand phalanges as indicators of sex among individuals from the Chiang Mai province of Thailand. The sample ranged in age from 19 to 93 years. Six measurements were taken on each proximal phalanx: maximum length, medio-lateral base width, antero-posterior base height, medio-lateral head width, antero-posterior head height and maximum mid-shaft diameter. The measurements were then subjected to ROC analysis as well as binary logistic regression to assess the relative correct allocation accuracy for each bone, and for different combinations of measurements from each bone. All proximal phalanges from both sides exhibited greater than 87% correct allocation accuracy for at least one logistic regression equation that included only two or three measurements. When the sample was limited to individuals with no missing measurements (n=209) in any of the phalanges, the most accurate equations for each proximal phalanx ranged from 87.6% to 92.3%, with the most accurate equation based on two measurements from the left 1st proximal phalanx, and the next most accurate from three measurements of the left 2nd proximal phalanx. The results suggest that proximal phalanges produce better allocation accuracies than metacarpals among modern individuals from Thailand. © 2013 Elsevier Ireland Ltd.