A relation exists between colloid osmotic pressure and serum total protein concentration; equations describing this relation have been used to determine a calculated value for colloid osmotic pressure. However, the relation between total protein concentration and colloid osmotic pressure is altered by the method used to measure protein and by changes in the ratio of concentrations of albumin (A) to globulin (G). We developed nomograms for estimating colloid osmotic pressure from A and G concentrations, using samples obtained from clinically normal animals and compared the accuracy of these nomograms with that of previously described equations relating colloid osmotic pressure to total protein concentration. For comparison, serum samples from canine (n = 106), equine (n = 79), feline (n = 24), and bovine (n = 27) patients admitted to the University of Georgia Veterinary Medical Teaching Hospital were used. Results indicated that nomograms based on protein concentration estimated by a refractometer generally were the least reliable. Although predictive nomograms, using total protein concentration determined by the biuret method, provided better results for serum samples, there was considerable variation between measured and calculated values for colloid osmotic pressure in all species studied. Calculated values for colloid osmotic pressure derived from A and G concentrations were most closely related to measured values for colloid osmotic pressure in dogs, horses, and cats. However, calculated values for colloid osmotic pressure differed from measured values by as much as 5 mm of Hg for some samples by each of the methods of estimation. These results indicate that, although calculated values for colloid osmotic pressure may be most accurate when variations in the A-to-G ratio are accounted for in the nomogram, none of the calculation methods provided a consistently accurate estimate of colloid osmotic pressure.