As the capabilities of automated license plate recognition (ALPR) devices increase, their applicability to measure individual travel times becomes appealing and the question of sizing the total number of devices becomes an important one. ALPR cameras have been deployed along a road of about 50 km in the French Alps. This road is highly congested, and users experience travel times of 35 to 120 min. Traffic managers plan to inform road users of their travel times on variable message signs. Slicing the complete journey into small sections (between two cameras) makes it possible to be more reactive to sudden changes in travel times. This is especially true if those changes are due to a congestion occurring at the beginning or in the middle of the journey. But increasing the total number of cameras is not cost-effective and must be avoided as much as possible. This paper presents the formulation and results of a simplified model for studying the impact of the total number of cameras on the precision of displayed travel times. The model was kept as simple as possible. Maximum use was made of the small amounts of data available. Comparison with on-site preliminary results proves that the model gives robust results.