Steady-state and time-resolved emission spectroscopy was applied to a study of the binary and ternary complexes of pure E. coli purine nucleoside phosphorylase (PNP) with phosphate (Pi; a substrate) and a close non-substrate analogue (sulfate; SA). The quenching of enzyme fluorescence by Pi was bimodal, best described by two modified Stem-Volmer equations fitted independently for "low" (below 0.5 mM Pi) and "high" (above 0.5 mM Pi) ligand concentrations. At Pi > 0.5 mM, binding is characterized by a fortyfold higher dissociation constant (Kd2 = 1.12 +/- 0.10 mM), i.e. by a lower affinity for phosphate, with a sevenfold lower quenching constant and 1.6-fold higher accessibility. By contrast, the binding of SA, and the resultant fluorescence quenching, was unimodal, with Kd = 1.36 +/- 0.07 mM, comparable to the Kd for Pi at "high" Pi, with a total binding capacity of one sulfate or phosphate group per enzyme subunit. SA proved to be a competitive inhibitor of phosphorolysis with Ki = 1.2 +/- 0.2 mM vs. Pi, hence similar to its Kd. SA at a concentration of 5 mM did not affect the Pi affinity at Pi < 0.5 mM, but led to a reduced affinity and twofold higher Pi binding capacities at Pi > 0.5 mM. The resultant fluorescence quenching by Pi decreased at 5 mM SA, with lower Stern-Volmer constant (KSV) and fractional accessibility (fa) values. Increasing concentrations of Pi reduced the enzyme affinity for SA, characterized by a higher Kd. The Hill model showed negative cooperative binding of Pi in the absence and presence of 5 mM SA with Hill coefficients h = 0.60 +/- 0.01 and h = 0.83 +/- 0.07, respectively. SA exhibited non-cooperative binding in the absence of Pi (h = 1.08 +/- 0.01) and negative cooperative binding in the presence of Pi (h < 1). PNP fluorescence decays were best fitted to a sum of two exponentials, with an average lifetime of 2.40 +/- 0.14 ns unchanged on interaction with quenching ligands, and pointing to static quenching. The overall results are relevant to the properties of PNP from various sources, in particular to the design of potent bisubstrate analogue inhibitors.