Two microbiological assay techniques for estimation of biologically active isoniazid concentrations in human serum were compared : the vertical diffusion method and a large plate method similar to that commonly used for the antibiotics. Four hundred seventy four tests were carried out on the serum from 133 patients. Parallel titrations were determined by two biologists on the same serum samples, with the same standard solutions, by the two methods performed rapidly after the collection of specimens and after they have been stored in the frozen state during one week and were compared. It was found that the variability obtained in these conditions was 26 per cent for the vertical diffusion method and 14 per cent for the large plate method. The correlation coefficients were 0,87 and 0,95 respectively. Also it will be noted that the large plate method is faster, of lower cost and require smaller volumes of serum. However, it also failed to detect levels lower than 0,5 and upper than 2 micrograms/ml, the standard curve is not stable and must be run with each assay and it is also subject to interference by other antibiotics.