The rate of change in resting metabolic rate (RMR) as a result of a temperature increase of 10 °C is termed the temperature coefficient (Q10), which is often used to predict how an organism’s total MR will change with temperature. However, this method neglects a potentially key component of MR; changes in activity level (and thus activity MR; AMR) with temperature may significantly alter the relationship between MR and temperature. The present study seeks to describe how thermal effects on total MR estimated from RMR–temperature measurements can be misleading when the contribution of activity to total MR is neglected. A simple conceptual framework illustrates that since the relationship between activity levels and temperature can be different to the relationship between RMR and temperature, a consistent relationship between RMR and total MR cannot be assumed. Thus the thermal effect on total MR can be considerably different to the thermal effect on RMR. Simultaneously measured MR and activity from three ectotherm species with differing behavioural and physiological ecologies were used to empirically examine how changes in temperature drive changes in RMR, activity level, AMR and the Q10 of MR. These species exhibited varied activity– and MR–temperature relationships, underlining the difficulty in predicting thermal influences on activity levels and total MR. These data support a model showing that thermal effects on total MR will deviate from predictions based solely on RMR; this deviation will depend upon the difference in Q10 between AMR and RMR, and the relative contribution of AMR to total MR. To develop mechanistic, predictive models for species’ metabolic responses to temperature changes, empirical information about the relationships between activity levels, MR and temperature, such as reported here, is required. This will supersede predictions based on RMR alone.