Although the impact of hydrodynamic conditions in a flotation cell is often evaluated by correlating impeller tip speed with bubble size, the literature reports inconsistent results, some showing a reduction in bubble Sauter mean diameter (d(32)) with increasing impeller speed, others showing little to no effect. A review of these results indicate that cell size may be a factor where small laboratory-scale cells, smaller than 50 L, tend to support the correlation while larger machines do not. This paper demonstrates an alternative approach using the average turbulent kinetic energy (TKE) in place of the impeller speed. Results were obtained using two cells with the same geometry but different size, 5 L and 60 L Bubble size (d(32)) was measured using the Anglo Platinum Bubble Sizer. Local velocity and velocity fluctuation were measured using a constant temperature anemometer to estimate the average turbulent kinetic energy (TKE). The effect of impeller tip speed on d(32) and TKE as a function of air rate was determined. Combining the results for the two cells showed that d(32) initially decreased with increasing TKE to become constant above a critical TKE. The TKE region below critical was associated largely with the 5 L cell and the region above critical more associated with the 60 L cell. The inconsistent data in the literature has been explained by introducing the concept of the critical TKE and it has been confirmed that the reported effect of increasing impeller speed may have its origin in the size of the cells tested: laboratory scale cells showing an effect on reducing bubble size as TKE is below critical while large and industrial scale cells may not as TKE is above critical. (C) 2013 Elsevier Ltd. All rights reserved.