An experiment was carried out to monitor food temperature changes with time in two insulated boxes equipped with an ice pack. The first experiment was carried out in a test room under well controlled ambient temperature and with a box of reinforce lateral insulation. The second experiment was in real use condition and with 2 different boxes. Two models were developed to predict product temperature changes until the ice is completely melted: an analytical 1-D and a 3-D model. The 1D model predictions are in good agreement for the first experiment. But in real use condition, the 1D model underestimates the temperature evolution, while the 3-D model gives better prediction. A comparison of the advantages and disadvantages of these two models was performed. The 1-D approach enables prediction of the main parameters (warmest temperature, ice melting time). A correction factor was proposed to improve the prediction precision using the 1-D model.