The purpose of the present study was to analyze longitudinal changes in 3,000 m running performance and the relationship with selected physiological parameters. Eighteen well-trained male middle-distance runners were measured six times (x3 per year) throughout two consecutive competitive seasons. The following parameters were measured on each occasion: maximal oxygen uptake (VO₂max), running economy (RE), velocity at maximal oxygen uptake (vVO₂max), velocity at 4mmol L^-1 blood lactate concentration (V4), and performance velocity (km·h^-1) in 3,000 m time trials. Values ranged from 19.59 to 20.16 km·h^-1, running performance; 197 to 207 mL·kg^-1·km^-1. RE; 17.2 to 17.7 km·h^-1, V4; 67.1 to 72.5 mL·kg^-1·min^-1, VO₂max; and 19.8 to 20.2 km·h^-1, vVO₂max. A hierarchical linear model was used to quantify longitudinal relationships between running performance and selected physiological variables. Running performance decreased significantly over time, between each time point the decrease in running velocity was 0.06 km·h^-1. The variables that significantly explained performance changes were V4 and vVO₂max. Also, vVO₂max and V4 were the measures most strongly correlated with performance and can be used to predict 3,000 m race velocity. The best prediction formula for 3,000 m running performance was: y = 0.646 + 0.626x + 0.416z (R²=0.85); where y = V3,000 m velocity (km·h^-1), x = V4 (km·h^-1) and z = vVO₂max (km·h^-1). The high predictive power of vVO₂max and V4 suggest that both coaches and athletes should give attention to improving these two physiological variables, in order to improve running performance.