A study of nerve impulse generation in ganglion cells of the tiger salamander retina is carried out through a combination of experimental and analytic approaches, including computer simulations based on a single- compartment model. Whole cell recordings from ganglion cells were obtained using a superfused retina-eyecup preparation and studied with pharmacological and electrophysiological techniques, including phase plot analysis. Experimental efforts were guided by computer simulation studies of an excitability model consisting of five voltage- or ion-gated channels, which were identified from earlier voltage-clamp data. The ion channels include sodium, calcium, and three types of potassium channels, namely the A type (I(KA)), Ca-activated potassium (I(K,Ca)), and the delayed rectifier (I(K)). A leakage channel was included to preserve input resistance continuity between model and experiment. Ion channel densities of Na and Ca currents (I(Na) and I(Ca)) for the single-compartment model were independently determined from phase plot analysis. The I(K) and I(K,A) current densities were determined from the measured width of impulses. The I(K,Ca) was modeled to respond to Ca influx, and a variable-rate Ca-sequestering mechanism was implemented to remove cytoplasmic calcium. Impulse frequency increases when either I(Ca) or I(K,Ca) is eliminated from the model or blocked pharmacologically in whole cell recording experiments. Faithful simulations of experimental data show that the ionic currents may be grouped into small (I(K,Ca), leakage, and stimulus), and large (I(Na), I(K), I(A), I(Ca)) on the basis of their peak magnitudes throughout the impulse train. This division of the currents is reflected in their function of controlling the interspike interval (small currents) and impulse generation (large currents). Although the single-compartmental model is qualitatively successful in simulating impulse frequency behavior and its controlling mechanisms, limitations were found that specifically suggest the need to include morphological details. The spike train analysis points to a role for electrotonic currents in the control of the duration of the interspike intervals, which can be compensated by prolonged activation of g(K,Ca) in the single-compartment model. A detailed, multicompartmental model of the ganglion cell is presented in the companion paper.