This paper examines the economics of traveler information from probe vehicles to understand how many probes are needed to provide useful information, and how that probe information might be supplied to travelers. Probes differ from permanently installed roadway detection devices both because they provide information that is less current and because an information system centered on this technology can be organized in the form of private clubs rather than a government agency. This paper estimates travel time associated with various shares of probes among the fleet by simulating different levels of probes, information subscription, and congestion. It examines the travel time savings under both recurring and non-recurring congestion. When there is non-recurring congestion, a low frequency of probes is sufficient to detect the incident and enable information consumers to choose alternates. However, smoothing the stochastic nature of traffic under recurring congestion requires a relatively high share of probes (up to one-third of the fleet), depending on the level of congestion.