Abstract. The Lewis Glacier on Mt. Kenya is one of the best studied tropical glaciers and has experienced considerable retreat since a maximum extent in the late 19th century (L19). From distributed mass and energy balance modelling, this study evaluates the current sensitivity of the surface mass and energy balance to climatic drivers, explores climate conditions under which the L19 maximum extent might have been sustained, and discusses the potential for using the glacier retreat to quantify climate change. Multi-year meteorological measurements at 4828 m provide data for input, optimization, and evaluation of a spatially distributed glacier mass balance model to quantify the exchanges of energy and mass at the glacier-atmosphere interface. Currently the glacier loses mass due to the imbalance between insufficient accumulation and enhanced melt, because radiative energy gains cannot be compensated by turbulent energy sinks. Exchanging model input data with synthetic climate scenarios, which were sampled from the meteorological measurements and account for coupled climatic variable perturbations, reveals that the current mass balance is most sensitive to changes in atmospheric moisture (via its impact on solid precipitation, cloudiness, and surface albedo). Positive mass balances result from scenarios with an increase of annual (seasonal) accumulation of 30 % (100 %), compared to values observed today, without significant changes in air temperature required. Scenarios with lower air temperatures are drier and associated with lower accumulation and increased net radiation due to reduced cloudiness and albedo. If the scenarios currently producing positive mass balances are applied to the L19 extent, negative mass balances are the result, meaning that the conditions required to sustain the glacier in its L19 extent are not reflected in today's meteorological observations using model parameters optimized for the present-day glacier. Alternatively, a balanced mass budget for the L19 extent can be achieved by changing both climate and optimized gradients (used to extrapolate the meteorological measurements over the glacier) in a manner that implies a distinctly different coupling between the glacier's local surface-air layer and its surrounding boundary layer. This result underlines the difficulty of deriving palaeoclimates for larger glacier extents on the basis of modern measurements of small glaciers.