This study aimed at investigating the relative roles of the duration versus intensity of exercise on the metabolic adaptations in vastus lateralis to short-term (10 day) aerobic-based cycle training. Healthy males with a peak aerobic power ( $$ {\dot{\text{V}}\text{O}}_{{ 2 {\text{peak}}}} $$ ) of 46.0 ± 2.0 ml kg−1 min−1 were assigned to either a 30-min (n = 7) or a 60-min (n = 8) duration performed at two different intensities (with order randomly assigned), namely moderate (M) and heavy (H), corresponding to 70 and 86 % $$ {\dot{\text{V}}\text{O}}_{{ 2 {\text{peak}}}} $$ , respectively. No change (P > 0.05) in $$ {\dot{\text{V}}\text{O}}_{{ 2 {\text{peak}}}} $$ was observed regardless of the training program. Based on the metabolic responses to prolonged exercise (60 % $$ {\dot{\text{V}}\text{O}}_{{ 2 {\text{peak}}}} $$ ), both M and H and 30 and 60 min protocols displayed less of a decrease (P < 0.05) in phosphocreatine (PCr) and glycogen (Glyc) and less of an increase (P < 0.05) in free adenosine diphosphate (ADPf), free adenosine monophosphate (AMPf), inosine monophosphate (IMP) and lactate (La). Training for 60 min compared with 30 min resulted in a greater protection (P < 0.05) of ADPf, AMPf, PCr and Glyc during exercise, effects that were not displayed between M and H. The reduction in both $$ {\dot{\text{V}}\text{O}}_{ 2} $$ and RER (P < 0.05) observed during submaximal exercise did not depend on training program specifics. These findings indicate that in conjunction with our earlier study (Green et al., Eur J Appl Physiol, 2012b), a threshold exists for duration rather than intensity of aerobic exercise to induce a greater training impact in reducing metabolic strain.