A study of the indentation size effect (ISE) in aluminum and alpha brass is presented. The study employs rate effects to examine the fundamental mechanisms responsible for the ISE. These rate effects are characterized in terms of the rate sensitivity of the hardness, partial derivativeH/partial derivative ln (epsilon)over-dot(eff), where H is the hardness and (epsilon)over-dot(eff) is an effective strain rate in the plastic volume beneath the indenter. partial derivativeH/partial derivative ln (epsilon)over-dot(eff) can be measured using indentation creep, load relaxation, or rate change experiments. The activation volume V*, calculated based on partial derivativeH/partial derivative ln (epsilon)over-dot(eff) which can traditionally be used to compare rate sensitivity data from a hardness test to conventional uniaxial testing, is calculated. Using materials with different stacking fault energy and specimens with different levels of work hardening, we demonstrate how increasing the dislocation density affects V*; these effects may be taken as a kinetic signature of dislocation strengthening mechanisms. We noticed both H and partial derivativeH/partial derivative ln (epsilon)over-dot(eff) (V*) exhibit an ISE. The course of V* vs. H as a result of the ISE is consistent with the course of testing specimens with different level of work hardening. This result was observed in both materials. This suggests that a dislocation mechanism is responsible for the ISE. When the results are fitted to a strain gradient plasticity model, the data at deep indents (microhardness and large nanoindentation) exhibit a straight-line behavior closely identical to literature data. However, for shallow indents (nanoindentation data), the slope of the line severely changes, decreasing by a factor of 10, resulting in a "bilinear behavior". (C) 2003 Elsevier Science Ltd. All rights reserved.