Freezing of gait (FOG) is one of the most debilitating symptoms experienced by people with Parkinson's Disease. Cueing strategies have been shown to help people overcome FOG. Current systems do not adequately capture the complexity of FOG and often struggle to differentiate it from similar movement patterns, such as voluntary stops. FOG is also associated with complex psychological and physiological processes, suggesting that changes in autonomic nervous system activity may accompany freezing episodes. Including sensors that measure autonomic changes in the nervous system could therefore complement existing approaches to FOG monitoring. Previous studies have shown that skin conductance (SC) and its features change in temporal proximity to FOG episodes. We analyzed the SC of 19 participants with self-reported daily FOG episodes in a semi-free-living environment. The experiments were performed in ON and OFF medication states. We compared four features extracted from the SC of freezing episodes and pre-freezing to voluntary stops, walking, and their interaction with medication using generalized linear mixed models. We found that the median, tonic level, and area under the curve of the phasic component differed significantly across motor conditions and temporal states, particularly when comparing FOG with walking, and were modulated by medication state. No consistent differences were observed between freezing and voluntary stopping across all features. Our results demonstrate that skin conductance captures autonomic differences related to motor context, medication state, and temporal proximity to freezing, including changes that occur shortly before a freeze. Incorporating these physiological markers into wearable technology could support multimodal, medication-aware approaches to FOG monitoring and cueing, potentially improving the daily lives of individuals with FOG.