Measurement of the energy expenditure is crucial to understand the biophysics of any kind of human locomotion. Despite the promising application of inertial measurement unit (IMU) for quantification of the energy expenditure during human on-land activities, it has never been deployed before to calculate the aquatic activities energy expenditure. Wearable IMUs were used in our study to capture biomechanically interpretable descriptors of swimming. These descriptors were fed as the input to a Bayesian linear model for estimation of the energy expenditure. To enhance generalization capacity of the estimation, a non-linear adjustment of the Bayesian model was devised using swimmer’s anthropometric parameters. We used a set of four waterproofed IMUs worn on forearms, sacrum and right shank of eighteen swimmers to extract the main spatio-temporal determinants of the front-crawl energy expenditure. The swimmers performed three 300-m trials at 70%, 80% and 90% of their 400-m personal best time. At the end of each 300-m the reference value of energy expenditure was measured based on indirect calorimetry and blood lactate concentration. The assessment of the proposed model on the test data shows a strong association between the estimated and reference energy expenditure (Spearman’s rho = 0.93, p-value <0.001) and a high relative precision of 9.4%. The backward elimination of model parameters with minimum RMS error criterion shows that by excluding the features extracted from forearm sensors i.e. using only two IMUs we can still achieve an accuracy of 0.9±11.3%.