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 (R(2)=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. Key pointsV4 and vVO2max are the most important physiological variables to explain longitudinal changes in 3000 m running performance;3000 m running performance prediction is better if one uses both V4 and vVO2max in the same formula: y = 0.646 + 0.626x + 0.416z; R(2)=0.85, where y is the Vrace (km/h), x is V4 (km/h) and z is vVO2max (km/h).The V4 and vVO2max can be used for training control purposes.