Mobile internet Quality of Service is a very important thing for service providers and customers, establishing an objective di erence between the service that the costumers pay for and what they actually get. This study proposes a method to forecast mobile internet QoS, using time series analysis and based on metrics obtained from passive measurements performed by Android mobile devices like signal strength, amount of data tra c and disconnection events. Also, the method considers external information about the environment that can have an e ect on mobile internet QoS.