Sodium benzyl penicillin, procaine penicillin and benethamine penicillin were applied into the bovine conjunctival sac as an aqueous solution or in ointment form in order to study the concentration‐time profiles. The series of treatments was repeated in five animals in a random sequence. Penicillin concentration in conjunctival sac fluid (CF) was determined using the agar‐well‐diffusion assay technique. The data obtained were transformed to linear regression slopes. Similarity of the slopes within treatments (in five different eyes) enabled the construction of four common lines by co‐variance analysis to^represent each treatment. The regression coefficients of the four common lines were then compared to study the difference between treatments. Topical application of 5,000 iu sodium benzylpenicillin in aqueous solution at a concentration isotonic with 0.9% saline, produced a duration of therapeutic concentration (DTC) in CF of 12.6 ± 1.5 h. When the same salt or other less water‐soluble ones were formulated at the same concentration in an ointment base, the DTC was significantly prolonged. For all treatments, peak concentrations in CF were recorded at the first sampling and ranged between 7 iu/ml and 14 iu/ml. Sodium benzylpenicillin or procaine penicillin, both in the ointment base, produced DTCs of 38.8 ± 2.1 h and 37 ± 4.0 h, respectively, while the ointment formulation of benethamine penicillin produced a DTC of 56 ± 4.5 h. The prolonged duration observed in the eye ointments can be partly accounted for by the viscous nature of the base. Other differences may be dependent on relative water solubility of each penicillin product and complexity of the surface mucos