Electron spin resonance has the potential to be a powerful analytical tool. The theoretical l.o.d. is ca. 10[raised -9]M, and the esr signal parameters can be related to the spin concentration over a wide range (10[raised -2] to 10[raised -9]M) with adjustment of only instrument receiver gain. Previous analytical studies have not taken full advantage of this potential and have not well characterized the precision with which analytical esr measurements can be made. Sample preparation has been a major obstacle. This study reports the identification of several new esr sample handling techniques, along with the characterization and evaluation (both experimentally and statistically) of parameters affecting the sensitivity and measurement of the esr signal obtained. Photochemical titration employing esr as an end point indicator can be done with 2.5% precision but the extreme range of concentrations which esr can detect is sacrificed. Resonance signal intensity parameters were evaluated and found to be linear with spin concentration, if spin environment is carefully controlled; sample matrix effects seriously distort the signal. Ion-exchange of paramagnetic transition metal paramagnetic ions from solution onto resins has been found advantageous (viz, solid state, low dielectric). Quantitative esr studies of first row transition metal ions bound to ion-exchange resins have demonstrated a number of experimental realities. Variables such as solution conditions (oxidation state) stability, complex ion formation and pH dependence of these equilibria) and the nature of the resin (strong or weak exchanger and type of polymeric resin backbone) limits the choices of ion/ligand/resin combinations and each must be considered to obtain a sensitive esr signal. For example, we find that the background signal due to the polymeric resin backbone determines the limit of detection for the combinations studied. Selection of an ion/ligand combination for a given d electron configuration is facilitated by Tanabe-Sugano theory which can be used to predict whether spin relaxation will render a signal which is measurable. Good agreement for a variety of Fe(III) and Co(II) ligand combinations has been obtained. By combining all of the factors mentioned above, we can list the necessary restrictions that analytically define the most sensitive combinations of metal ion/ligand/resin which are consistent with simple sample preparation. The effects of random errors in instrument parameter settings and in sample preparation techniques have been evaluated in terms of statistical reproducibility (rsd's of 4.3% and 3.3%). The ammine and o-phenanthroline complexes of Cu(II) exchanged on strong and weak resins were used to determine the upper limit of linearity (10[raised -3] and 10[raised -4]M) lower limit of detection (10[raised -6] and 10[raised -7]M), concentration range over which the determination may be conducted and statistical parameters (RSEE's of 8% for three orders of magnitude and 4% for two orders of magnitude). New sample preparation techniques of dry dilution and spin redistribution were developed. Dry dilution allows the effect of interspin interaction of spins from one bead to another to be diluted; while spin redistribution causes the effect of intraspin interaction of spins on a single bead with each other to be diminished. The upper limit of linearity is determined by these effects. Qualitative information obtained from esr spectra has been shown useful for studying resin loaded with Cu(II) which are used for ligand-exchange chromatography. The strength of analytical esr measurements is the wide concentration range available. The weakness of the measurement is in obtaining a sample matrix which provides a suitable relaxation time without altering the signal intensity in ways not conducive to analytical measurements.