Hourly observations of forest-atmosphere energy balance components are presented for the Morgan-Monroe State Forest, south-central Indiana, USA for the period March 1998 to December 2001, with particular emphasis on the storage heat flux (ΔQS). The sub-components of ΔQS showed strong diurnal variability, although ground heat flux dominated the seasonal change of storage heat flux. The annual storage heat balance for the three complete years showed a small but consistent deficit averaging 16.18 MJ m−2 a−1. On a per-hour basis this is small compared with potential error highlighted in sensitivity testing of a range of methodological issues. Under-sampling of biomass temperature yielded the greatest uncertainty. Energy balance closure, determined by linear regression, found turbulent fluxes to underestimate available energy by 28% (r2 = 0.92), with no statistically significant difference in closure found between years or between months. On a diurnal basis, weaker closure was associated with low friction velocity and low turbulent heat flux magnitude. Increasing turbulent flux block averaging time from 15 to 60 min resulted in a small improvement in closure at 60 min averages (2.5%). In addition, co-spectral corrections to the latent heat flux to account for long-tube damping effects generated a much larger improvement in closure (16%). The mean annual Bowen ratio was 0.59, ranged from 0.57 to 0.61 and inter-annual variability of all energy balance components over the period was very small (less than 6% departure from mean).