The Tianlai cylinder array is a pathfinder for developing and testing 21 cm intensity mapping techniques. In this paper, we use numerical simulation to assess how its measurement is affected by thermal noise and the errors in calibration and map-making process, and the error in the sky map reconstructed from a drift scan survey. Here we consider only the single frequency, unpolarized case. The beam is modeled by fitting to the electromagnetic simulation of the antenna, and the variations of the complex gains of the array elements are modeled by Gaussian processes. Mock visibility data are generated and run through our data processing pipeline. We find that the accuracy of the current calibration is limited primarily by the absolute calibration, where the error comes mainly from the approximation of a single dominating point source. We then studied the m-mode map-making with the help of Moore–Penrose inverse. We find that discarding modes with singular values smaller than a threshold could generate visible artifacts in the map. The impacts of the residue variation of the complex gain and thermal noise are also investigated. The thermal noise in the map varies with latitude, being minimum at the latitude passing through the zenith of the telescope. The angular power spectrum of the reconstructed map show that the current Tianlai cylinder pathfinder, which has a shorter maximum baseline length in the North–South direction, can measure modes up to l ≲ 2πbNS/λ ∼ 200 very well, but would lose a significant fraction of higher angular modes when noise is present. These results help us to identify the main limiting factors in our current array configuration and data analysis procedure, and suggest that the performance can be improved by reconfiguration of the array feed positions.