Abstract Many natural water transformations are mediated by reactive transients (RTs) too fleeting to measure directly. The concepts of molecular probe methods that can uniquely characterize these otherwise inaccessible species are critically reviewed in the context of environmental studies. Photochemically produced free radicals exemplify important RTs. A photooxidation scheme which characterizes the radical chemistry of seawater and emphasizes the reactions of oxygen species is presented. All the hypothesized radicals are measurable only as RTs with lifetimes < 10 ms; many undergo critical reactions on the 1 μs time-scale. New probe systems have been designed for use at sea and to explore and test the photooxidation hypothesis. These stable free radical probes have kinetic and thermodynamic advantages compared with conventional even-electron probes as a result of elementary chemical considerations. Their selectivity profiles match well the properties of the primary and secondary radical groups postulated by the hypothesis, and initial data demonstrate that these systems are sensitive, give oceanographically consistent results, and tentatively confirm or constrain the scheme. Organic nitroxide probes confirm the postulated fast radical + O 2 reactions, and can be used to separate and identify trace organic radicals in model systems and natural waters. The inorganic probe nitric oxide has been used extensively to measure total radical fluxes of ∼ 0.1–1 nM kg −1 min −1 of noon surface insolation in ‘blue’ waters (∼ 10–100 μM per calendar year), and has confirmed geochemically significant flux magnitudes. The fluxes of the superoxide ion-radical have been measured at sea using related 15N-labelled NO probes. Values approximately one-third of the total flux at the same sites confirm that this previously postulated species is a major component of the photochemical radical array, and qualitatively strengthen the hypothesis that this species is the source of photochemically produced HOOH. Testing the robustness of results from probe studies of the photooxidation hypothesis requires intercalibration methods and probe-independent validation experiments. Prospects for realizing these constraints are surveyed.