To investigate discrimination of firing signatures in the battlespace, blast, and flash signatures were collected from two large-caliber guns. Signatures included blast wave trajectories from high-speed (1600 Hz) imagery, visible and near-infrared (450 to 850 nm) spectra, and mid-wave infrared (1800 to 6000 cm−1) spectra. Thirty empirical and 12 phenomenological features were extracted from the signatures over 25 observations. Multiple discriminant analysis was used for feature selection and dimensionality reduction, and discrimination was based on Bayesian probability theory. When discriminating between three munitions configurations fired from a 152 mm howitzer, leave-one-out testing with the empirical features demonstrated classification accuracies of 92% to 96%. Phenomenological features (for example, blast energy, combustion and excitation temperatures, and species concentrations) accurately classified munitions less consistently (72% to 96%). The two most salient features for differentiating between two weapons (a single 120 mm tank firing and all 152 mm howitzer firings) were soot emissivity and H2O concentration. Empirical features consistently discriminated the weapons best when more than two features were used. Moderate to strong correlations (r2=0.5 to 1.0) were observed amongst blast, visible, and infrared features. A limited examination indicated that strongly correlated features may be substituted for one another.