The association between individual plant performance and seed yield variability within and between field pea crops was investigated. In 1988/89 six F8 genotypes with morphologically distinct characteristics were selected from a yield evaluation trial. Analysis of the individual plant performance within these crops indicated an association between low seed yields and the location and dispersion of plant harvest index (PHI) and plant weight (PWT) distributions. The analyses also showed there was a strong linear relationship between the seed weight (SWT) and PWT of the individual plants within each crop, and that the smallest plants tended to have the lowest PHI values. A series of 20 simulations was used to formalize the relationships between SWT, PWT and PHI values within a crop into a principal axis model (PAM). The PAM was based on a principal axis which represented the linear relationship between SWT and PWT, and an ellipse which represented the scatter of data points around this line. When the principal axis passed through the origin, the PHI of a plant was independent of its PWT and the mean PHI was equal to the gradient of the axis. However, when the principal axis had a negative intercept then the PHI was dependent on PWT and a MPW was calculated. In 1989/90 four genotypes were sown at five plant populations, ranging from 9 to 400 plants m⁻². Significant seed and biological yield differences were detected among genotypes at 225 and 400 plants m⁻². The plasticity of yield components was highlighted, with significant genotype by environment interactions detected for each yield component. No relationship was found between results for yield components from spaced plants and those found at higher plant populations. The two highest yielding genotypes (CLU and SLU) showed either greater stability or higher genotypic means for PHI than genotypes CVN and SVU. Despite significant skewness and kurtosis in the SWT, PWT, and PHI distributions from the crops in this experiment, the assumptions of the PAM held. The lower seed yield and increased variability in PHI values for genotype CVN were explained by its higher MPW and the positioning of the ellipse closer to the PWT axis intercept than in other genotypes. For genotype SVU, the lower seed yield and mean PHI values were explained by a lower slope for the principal axis. Both low yielding genotypes were originally classified as having vigorous seedling growth and this characteristic may be detrimental to crop yields. A method for selection of field pea genotypes based on the PAM is proposed. This method enables the identification of weak competitors as single plants, which may have an advantage over vigorous plants when grown in a crop situation.