A filter based near infrared (NIR) technique to evaluate Brix value, in a non destructive way, of Indian apples, along with apples from other countries, has been investigated here. Reflectance characteristics at ten NIR wavelengths were recorded by in-house developed probe for five Indian varieties and four imported varieties. A separate calibration and validation set for each variety was formed. Brix values of the calibration and validation sets of apples were evaluated with a refractometer from Bellingham and Stanley, Model 840. Principal component regression was used to generate calibration and prediction of Brix for the validation set. It was found that Brix values of all varieties of apples can be predicted with a standard error of prediction, 0.7322–1.7809, while root mean square error is also in the same range. It was further noticed that reduction in data recording wavelength leads to an increase in SEP and RMSE.