The M 7.0 Haiti earthquake of 12 January 2010 caused catastrophic damage and loss of life in the capital city of Port‐au‐Prince. The extent of the damage was primarily due to poor construction and high population density. The earthquake was recorded by only a single seismic instrument within Haiti, an educational seismometer that was neither bolted to the ground nor able to record strong motion on scale. The severity of near‐field mainshock ground motions, in Port‐au‐Prince and elsewhere, has thus remained unclear. We present a detailed, quantitative analysis of the marks left on a tile floor by an industrial battery rack that was displaced by the earthquake in the Canape Vert neighborhood in the southern Port‐au‐Prince metropolitan region. Results of this analysis, based on a recently developed formulation for predicted rigid body displacement caused by sinusoidal ground acceleration, indicate that mainshock shaking at Canape Vert was approximately , corresponding to a modified Mercalli intensity of VIII. Combining this result with the weak‐motion amplification factor estimated from aftershock recordings at the site as well as a general assessment of macroseismic effects, we estimate the peak acceleration to be for sites in central Port‐au‐Prince that experienced relatively moderate damage and where estimated weak‐motion site amplification is lower than that at the Canape Vert site. We also analyze a second case of documented rigid body displacement, at a location less than 2 km from the Canape Vert site, and estimate the peak acceleration to be approximately at this location. Our results illustrate how observations of rigid body horizontal displacement during earthquakes can be used to estimate peak ground acceleration in the absence of instrumental data.