Fatigue assessment and quantification are essential requirements to reduce the risks that occur as a consequence of a fatigued operator. The new wearable device technology offers an accurate measuring ability to one or more of fatigue-related biological data, which helps in quantifying fatigue levels in real-life environments. This paper presents a new heart rate variability (HRV) based operator-fatigue analysis and classification method using low-cost wearable devices. HRV that is considered a robust fatigue metric is measured by several wearable devices including a chest-strap heart monitor and a wrist watch that measures heart rate, skin temperature and skin conductivity. The data collected from real subjects are used to create a training dataset for fatigue analysis and classification. Two supervised machine-learning algorithms based on multi-layer neural network and support vector machine are developed and implemented to identify the alertness/fatigue states of the operator. Performance of the developed classifiers demonstrated high alertness/fatigue prediction accuracy. Such findings proved that the proposed analysis and classification method is valid and practically applicable.