The imposition of a minimum total soluble solids (TSS) value as a quality standard for orange-flesh netted melon fruit (Cucumis melo L. reticulatus group) requires either a batch sampling procedure (i.e. the estimation of the mean and standard deviation of a population), or the individual assessment of fruit [e.g. using a non-destructive procedure such as near infrared (NIR) spectroscopy]. Several potential limitations to the NIR assessment of fruit, including the variation in TSS within fruit and the effect of fruit storage conditions on the robustness of calibration models, were considered in this study. Outer mesocarp TSS was 3 TSS units higher at the stylar end of the fruit compared with the stem end, and the TSS of inner mesocarp was higher than outer tissue and more uniform across spatial positions. The linear relationship between the outer 10 mm and the subsequent middle 10 mm of tissue varied with fruit maturity [e.g. 42 days before harvest (DBH), r ² = 0.8; 13 DBH, r ² = 0.4; 0 DBH, r ² = 0.7], and with cultivars (at fruit maturity, Eastern Star 2001 r ² = 0.88; Malibu 2001 r ² = 0.59). This relationship notably affected NIR calibration performance (e.g. based on inner mesocarp TSS; R _c ² = 0.80, root mean standard error of cross-validation (RMSECV) = 0.65, and R _c ² = 0.41, RMSECV = 0.88 for mature Eastern Star and Malibu fruit, respectively). Cold storage of fruit (0–14 days at 5°C) did not affect NIR model performance. Model performance was equivalent when based on either that part of the fruit in contact with the ground or equatorial positions; however, it was improved when based on the stylar end of the fruit.