We recently developed a single photon radioluminescence (SPR) technique to measure submicroscopic distances in biological samples [Bicknese et al., and Shahrokh et al., Biophys. J., 63 (1992) 1256–1279]. SPR arises from the excitation of a fluorophore by the energy deposited from a slowing beta decay electron. The purpose of this study was to detect 3H 2O molecules near tryptophan residues in proteins by tryptophan SPR. To detect small SPR signals, a sample compartment with reflective ellipsoidal optics was constructed, and amplified signals from a cooled photomultiplier were resolved by pulse-height analysis. A Monte Carlo calculation was carried out to quantify the relationship between SPR signal and 3H 2O-tryptophan proximity. Measurements of tryptophan SPR were made on aqueous tryptophan; dissolved melittin (containing a single tryptophan); native and denatured aldolase; dissolved aldolase, monellin, and human serum albumin; and the integral membrane proteins CHIP28 (containing a putative aqueous pore) and MIP26 using 3H 2O or the aqueous-phase probe 3H-3- O-methylglucose (OMG). After subtraction of a Bremsstrahlung background signal, the SPR signal from aqueous tryptophan (cps · μCi −1 μmol −1 ± SE) was 8.6 ± 0.2 with 3H 2O and 7.8 ± 0.3 with 3HOMG ( n = 8). With 3H 2O as donor, the SPR signal (cps · μCi −1 μmol −1) was 9.0 ± 0.3 for monomeric melittin in low salt (tryptophan exposed) and 4.6 ± 0.8 ( n = 9) for tetrameric melittin in high salt (tryptophans buried away from aqueous solution). The ratio of SPR signal obtained for aldolase under denaturing conditions of 8 M urea (fluorophores exposed) versus non-denaturing buffer (fluorophores buried) was 1.53 ± 0.07 ( n = 6). Ratios of SPR signals normalized to fluorescence intensities for monellin, aldolase, and human serum albumin, relative to that for d-tryptophan, were 1.42, 1.09, and 1.04, indicating that the cross-section for excitation of fluorophores in proteins is greater than that for tryptophan in solution. For the CHIP28 and MIP26 proteins in membranes, the ratio of SPR signal obtained with 3H 2O versus 3HOMG was 1.35 ± 0.13 (CHIP28, n = 5) and 0.99 ± 0.02 (MIP26). These data are consistent with the existence of an aqueous channel through CHIP28 that excludes small solutes. We conclude that tryptophan radioluminescence in proteins is measurable and provides unique information about the presence of local aqueous compartments.