Thirty-seven compounds, including 17 biguanides, 6 carbamimidates, and 14 bisbiguanides, were evaluated for potential antiplaque activity by measuring their minimum inhibitory concentrations [MIC (M)] against Streptococcus mutans no. 6715. Linear regression analysis was conducted with the log 1/MIC (M) values and log P, pi, sigma, and MR. The best correlation for the biguanides (r2 = 0.92) was obtained with log P and (log P)2. When the biguanides were included with the carbamimidates, essentially the same correlation (r2 = 0.91) was obtained with log P and (log P)2. The best correlation for the bisbiguanides (r2 = 0.70) was also obtained with log P and (log P)2. Use of an indicator variable (I) for the bisgiguanides allowed all three groups to be included in one equation, which accounted for over 87% of the variance in the data for inhibition of bacterial growth. These results from the classical parabolic model were also compared with those from the recently developed bilinear model.