The variable aggregation of porphyrins such as Hp and Hpd introduces uncertainties and errors into attempts to measure their binding to proteins. Methods such as dialysis, ultrafiltration and gel chromatography, so frequently used, proved to be unreliable when applied to the binding of Hp to serum albumin. Quenching of tryptophan fluorescence will only occur at porphyrin binding sites which are closely situated to the tryptophan residue (1.7 nm). Porphyrin bound to more distant sites may not be included in this analytical procedure which must therefore be applied with reserve. In the present work, photofrin II (PII) was shown to consist of large aggregates > 20000–30000 M r, solutions of which did not disaggregate on dilution down to 1 μmol/1. Addition of albumin resulted in a change in the absorption spectrum of PII. Thus, it was assumed that measurements of differential absorption gave the proportion of free-to-bound PII when serum albumin was added in graded amounts to its solution. By applying suitable calculations to the data, an association constant of 0.3 l/ μmol ± 30% was deducted. Hill plots of the binding data were linear with slopes close to unity. Experimentally determined uptake of PII by NHIK 3025 cells from solutions containing different amounts of HSA showed that the amount bound to the cells was proportional to the free PII. The kinetics of quenching of tryptophan fluorescence in HSA by PII indicates that there is one main porphyrin-binding site affecting this fluorescence. This binding site seems to have a slightly higher affinity for PII than the remaining sites. Up to 8 porphyrin rings of PII can be bound to an HSA molecule.