The official first action method for carbadox in swine feed, 42.C01-42.C04, was modified in 2 respects. First, the samples were leached overnight at room temperature instead of boiled for 1 hr. This change avoided problems with overheating and excessive evaporation. Second, the dilution scheme for samples spiked with carbadox standard solution was changed to give absorbance values that were within the optimum working range of all types of spectrophotometers. The modified procedure was collaboratively studied by 21 laboratories. The repeatability standard deviation (sigma0) and reproducibility standard deviation (sigmax) were sigma0 = 0.00029% and sigmax = 0.00056% (8.9% of grand mean) for feeds containing 0.00617% carbadox; and sigma = 0.0012% and sigmax = 0.0019% (9.3% of grand mean) for feeds containing 0.0198% carbadox. The between-laboratory variance ratio was significant for feeds containing 0.0198% carbadox. The mean per cent of intent values for feeds containing 0.00617% carbadox and 0.0198% carbadox were 102% and 104%, respectively. In general, the statistical results were comparable to those previously obtained for the official first action method. Consequently, the modified procedure is not recommended as a replacement for the official first action method.