A finite element model is used to calculate temperature and velocity along a two-dimensional glacier flow line. The model incorporates automatic element mesh generation, interpolation from point input data to all boundary nodes, comprehensive output graphics, linkage between velocity and temperature under one control program to allow iteration between the two, and time stepping in discrete units. The model results are tested against analytical solutions (for ice temperatures and velocities in simple situations) and field data (ice thickness, velocity, temperature, and mass balance). Sensitivity analyses were conducted with respect to model resolution, changed time step interval, temperature gradient, and bedrock topography. Limitations and strengths of applying finite element techniques to the modeling of glaciers are discussed. Some problems are encountered with the technique and methods of computation (e.g., limitation of mesh size, discretizing of a continuous body, equation approximation). Others arise from an inability to handle complex ice conditions (e.g., basal sliding and non sliding/sliding transition, cavitation, basal water). The inadequacy of glaciological data sets to provide sufficiently accurate and numerous boundary conditions and input values is a further major problem. Nevertheless finite element methods hold considerable promise for modeling studies of cold, creep-dominated ice masses where are marked gradients in temperature and velocity. These conditions are least well handled by conventional numerical techniques.