1. 1. In this paper we have studied the modified ninhydrin procedure of yemm and cocking, in an attempt to evaluate its applicability to the determination of amino acid-N in blood, urine and cerebrospinal fluid. To accomplish this certain, adjustments in the technique were required in the case of each of the various biological fluids. 2. 2. The method showed good reproducibility (coefficient of variation = 4.83) and good recovery (average of six = 100.0%) when applied to bovine serum ultrafiltrate. It was noted, however, that with increasing concentration of added amino acid, recovery of total amino acid tended to increase. 3. 3. Applied to the analysis of 31 specimens of normal serum or plasma, the results indicate a mean value of 5.15 mg % of amino acid-N, a standard deviation of ± 0.46 and a coefficient of variation of 8.95. Comparison with the method of folin-frame revealed no significant difference between the means of the two sets of determinations. The variation for the nmhydrin method applied to normal sera is about 1.86 times that found for bovine serum ultrafiltrate. 4. 4. Analysis of serum or plasma from an abnormal population ( N = 36) including eight cases of phenylpyruvic oligophrenia, gave a mean of 5.14 mg% of amino acid-N, a standard deviation of ± 0.91 and a coefficient of variation of 17.7; thus indicating considerably greater variation in the abnormal than in normal serum specimens. The deviations in the direction of the higher values is primarily due to the 8 cases of phenylpyruvic oligophrenia (P.P.O.). When studied as a group these revealed considerable homogeneity, i.e., a mean of 6.41 mg% of amino acid-N, a standard deviation of ± 0.30 and a coefficient of variation of 4.62. The mean of the normal ( N = 31) and the abnormal serum specimens (without P.P.O.) revealed a significant decrease due to the presence of a large number of cases with amino acid values below normal. 5. 5. A study of 13 specimens of normal urine and 16 abnormal specimens gave large values for both standard deviation and coefficient of variation indicating that the method is less reliable in its application to urine because of difficulties in collecting complete 24-h urine samples. The abnormal cases showed a higher mean value, and a greater range of values, the principle cause of the dispersion being due to the 6 cases of Wilson's disease. 6. 6. Application of the method to 14 abnormal specimens of spinal fluid, indicated considerable variation (C.V. of 22.8%) but much less than in the case of the normal urine samples ( C. V. = 37.0%) or the abnormal urine samples ( C. V. = 74.0%).