A sensitivity study to perturbations in the initial conditions in a simulated evolution of a baroclinic wave using the Weather Research and Forecasting (WRF) regional model is discussed in this paper. With the goal of analysing the impacts of these perturbations in the context of weather forecast and also with the aim of exploring the presence of preferred directions of growth in the errors, two cases were analysed: an ideal experiment using a case study available from the WRF set-up and a simulation of a real evolution of a mid-latitude cyclone over Southeastern South America (SESA) and the Atlantic Ocean. In the ideal experiment two spatial structures were considered for the perturbations, random and sinusoidal noise, while for the real experiment only sinusoidal structures were considered. These perturbations were then applied to the initial conditions of temperature and zonal wind. Additionally, simulations with an increment in atmospheric moisture were performed for the real experiment. It was found in both real and ideal cases that the temperature errors tend to organize in a preferable direction of growth, although this direction changes depending on the case considered. Results also show that perturbations in temperature field lead to larger error than perturbing the zonal wind. Finally, the increment of moisture did not produce significant changes in the distribution or intensity of the errors in the real case.