Human-computer/machine interface research benefits from accurate human arm models for stability analysis, control, and system design. The current study developed models for human arm dynamics and variability specific to stylus-based kinesthetic haptic interfaces. Data from nine human subjects (5 male, 4 female, ages 20-30) were collected using a three degree-of-freedom haptic device in the X, Y, and Z axes along with a range of grip forces (1-3N) for parametric system identification of the human arm and hand. Variability models that accounted for subject and grip force variation were also identified. The arm and hand model structure consisted of a third-order linear parametric transfer function that was paired with a previously derived second-order model for the haptic robot. The variability was modeled as multiplicative unstructured uncertainty using transfer functions. All of the model parameters were identified in the frequency domain and have force as input and position as output.