Cellular networks are becoming more common, this introduces new challenges in dealing with their error prone nature. To improve end-to-end performance when the first link in the connection is wireless, an adaptive timeout based on network conditions is constructed. Relevant network factors are identified by examining data collected by a device located in a vehicle moving around in southern Sweden. Channel Quality Indicator (CQI) is shown to be the primary predictor of errors in the connection. In our datasets, a CQI index of 2 is a very good predictor of an error prone state. The collected data is split into training and evaluation data, the training data is used to construct a model. An adaptive timeout mechanism which uses this model is proposed, the mechanism is shown to be superior in all tested cases in the dataset compared to the optimal static counterpart. Reducing timeouts allows for applications to make new decisions based on new information faster, increasing responsiveness and user satisfaction.