The amount of Short Message Service (SMS) spam is increasing. Various solutions to filter SMS spam on mobile phones have been proposed. Most of these use Text Classification techniques that consist of training, filtering, and updating processes. However, they require a computer or a large amount of SMS data in advance to filter SMS spam, especially for the training. This increases hardware maintenance and communication costs. Thus, we propose to filter SMS spam on independent mobile phones using Text Classification techniques. The training, filtering, and updating processes are performed on an independent mobile phone. The mobile phone has storage, memory and CPU limitations compared with a computer. As such, we apply a probabilistic Naïve Bayes classifier using word occurrences for screening because of its simplicity and fast performance. Our experiment on an Android mobile phone shows that it can filter SMS spam with reasonable accuracy, minimum storage consumption, and acceptable processing time without support from a computer or using a large amount of SMS data for training. Thus, we conclude that filtering SMS spam can be performed on independent mobile phones. We can reduce the number of word attributes by almost 50% without reducing accuracy significantly, using our usability-based approach. Copyright © 2012 John Wiley & Sons, Ltd.