Semi-arid regions are frequently subject to major temperature changes during a 24 h period, which may drastically affect greenhouse indoor climates. In order to improve energy management of these buildings, numerical tools have been developed to predict the evolution of the inside climatic conditions. However, most of the available models neither take account of the transmittivity variation through the day nor of differences between wall temperatures. In the present paper, a model for predicting the thermal and water behaviour inside an unheated agricultural green-house is presented. The energy balance method is applied to each element: cover, indoor air and soil surface. Specific modules have been developed to calculate heat transfer coefficients for the cover of the greenhouse as well as heat transfer through the subsoil. These modules have been integrated in the TRNSYS environment. Radiative transfers and view factors were also calculated. The simulations predict two main parameters under transient conditions: the indoor air temperature and the indoor humidity in response to the outside conditions. These parameters were validated with fair agreement from experiments conducted in a monospan greenhouse located in Batna (6.11° E, 35.33° N). Based upon the results of the simulations and the measurements it was also concluded that firstly, the transmittivity was not constant in time and varied with surface orientation; and secondly, vertical surface temperatures were different during the daytime while the temperature difference between roof surfaces remained insignificant. The evolution of humidity was not correctly reproduced by the model, probably because the effects of condensation and variation of soil water content were not properly included in the equations.