On the basis of the difference among values stated as normal by different authors to protein content of cerebrospinal fluid the present study was enterprised to verify whether these differences might be introduced by technical conditions. Four methods for protein determination were studied; two based on turbidimetric reactions and two on colorimetric ones. The first two were the trichloracetic acid method of Bossak, Rosenberg and Harris and the sulphosalycilic acid method of Denis and Ayer; the colorimetric methods were those of Dittebrandt (Weichselbaum biuret reagent) and of Macheboeuf and Rebeyrotte (nitric acid reagent). The calibration was conducted in a Coleman Junior spectrophotometer model 6-A with carefully prepared dilutions of pools of cerebrospinal fluids clear and colourless whose protein content was previously determined by the Kjeldahl procedure. Determination of protein content were made in several specimens of cerebrospinal fluid by each one of the methods and concentrations calculated through the calibration curves established. The protein content of all samples was previously evaluated by the Nissl method in an approximative way. The data obtained showed an influence of the effective instrument error on the normal taxes, specially on the colorimetric methods analised. This was accompanied by a rise in the mean values for each one. The data of twenty specimens whose protein content was estimated concomitantly by the four methods bring further evidence as to agree with this. The frequency of distribution in concentration zones of four series of specimens studied by each one of the methods showed a larger distribution in regions of higher protein concentration for methods with increased influence of effective instrument error. The present data give a way to understand the different taxes supported as normal in the literature whose range is sometimes too much extent, bringing difficulty in the evaluation of slight increasings in protein content which bare importance in the early diagnosis of some neurological affections.