Four manual micromethods for protein determination, two turbidimetric (trichloroacetic and sulfosalicylic acid-sodium sulfate) and two colorimetric (Lowry and Coomassie Brilliant Blue--sodium dodecyl sulfate, CBB-SDS) were used to compare the standard curves for total protein (0.30 to 3 g/L) produced with three reference materials: bovine serum albumin, human serum albumin, and diluted human serum. We measured the apparent protein content of a sample of pooled human cerebrospinal fluid by all four methods and with use of all three standards. The only reference material that gave similar results with all four methods was diluted human serum; the CBB-SDS was the only method that gave identical results with all three reference materials. We then measured the protein concentration of 28 individual cerebrospinal fluid samples by the four methods, with diluted human serum as standard. Results by all methods correlated well, but only the sulfosalicylic acid and the CBB-SDS methods gave equivalent results. We conclude that the choice of standard is more important than the method used. However, the CBB-SDS method may be the preferred method because it produced identical standard curves with all three protein standards.