AbstractWe test the hypothesis that limitations in the sky model used to calibrate an interferometric radio telescope, where the model contains extended radio sources, will generate bias in the Epoch of Reionisation power spectrum. The information contained in a calibration model about the spatial and spectral structure of an extended source is incomplete because a radio telescope cannot sample all Fourier components. Application of an incomplete sky model to calibration of Epoch of Reionisation data will imprint residual error in the data, which propagates forward to the Epoch of Reionisation power spectrum. This limited information is studied in the context of current and future planned instruments and surveys at Epoch of Reionisation frequencies, such as the Murchison Widefield Array (MWA), Giant Metrewave Radio Telescope and the Square Kilometre Array (SKA1-Low). For the MWA Epoch of Reionisation experiment, we find that both the additional short baselineuv-coverage of the compact Epoch of Reionisation array, and the additional long baselines provided by TGSS and planned MWA expansions, are required to obtain sufficient information on all relevant scales. For SKA1-Low, arrays with maximum baselines of 49 km and 65 km yield comparable performance at 50 MHz and 150 MHz, while 39 km, 14 km, and 4 km arrays yield degraded performance.