Land-terminating ice cliffs (LTICs) are common features of glacier margins that can contribute disproportionately to glacier ablation, yet direct observations of their boundary-layer structure and turbulent exchange are almost entirely lacking. Here, we analyse a unique 42.5-hour eddy-covariance dataset collected at a 25 m tall near-vertical ice cliff on the summit glaciers of Kilimanjaro, Tanzania, to investigate the microclimate, turbulence structure, and surface–atmosphere exchange associated with an LTIC. The measurements reveal a geometry-induced secondary circulation superimposed on the prevailing along-cliff flow, with air moving towards and ascending the cliff during daytime and reversing at night. Strong thermal contrasts between the warm volcanic ash forefield and the cold ice surface produced pronounced spatial heterogeneity in turbulence. While the upper cliff exhibited a coherent, wall-bounded turbulence regime consistent with locally generated exchange, the cliff foot was characterized by disrupted turbulence, counter-gradient transport, and evidence of flow separation associated with the cliff geometry. Using these observations, we evaluate the applicability of a classical bulk aerodynamic parameterization to a near-vertical glacier surface. Despite the unconventional geometry, the bulk method reproduced the observed sensible and latent heat fluxes well at the upper cliff with an RMSE of 5.2 and 3.3 W m^-2, respectively. At the cliff foot, the bulk method failed to capture the nighttime flux reversal, indicating that local turbulence structure, rather than surface slope alone, determines the validity of bulk parameterizations. Compared with previous measurements over Kilimanjaro's adjacent flat glacier surface, the LTIC exhibited substantially weaker latent heat fluxes, suggesting a shift from the sublimation-dominated regime characteristic of the plateau glaciers toward a comparatively more melt-dominated energy balance at the cliff. These observations provide the first field-based characterization of turbulence at a LTIC and point towards the idea that steep glacier margins can sustain coherent boundary-layer exchange despite their near-vertical geometry, with implications for understanding and modelling turbulent heat exchange over complex glacier terrain. Given the short duration and single-site nature of the observations, these findings should be regarded as a first case-study insight into LTIC boundary-layer processes rather than a general characterization of turbulent exchange at steep glacier margins.