Background: Muscle mass and physical function (PF) are common co-primary endpoints in cancer cachexia trials, but there is a lack of data on how these outcomes interact over time. The aim of this secondary analysis of data from a trial investigating multimodal intervention for cancer cachexia (ClinicalTrials.gov: NCT01419145) is to explore whether changes in muscle mass and PF are associated with weight loss and cachexia status at baseline. Methods: Secondary analysis was conducted using data from a phase II randomized controlled trial including 46 patients with stage III–IV non-small cell lung cancer (n = 26) or inoperable pancreatic cancer (n = 20) due to commence chemotherapy. Cachexia status at baseline was classified according to international consensus. Muscle mass (assessed using computed tomography (CT)) and PF outcomes, i.e., Karnofsky performance status (KPS), self-reported PF (self-PF), handgrip strength (HGS), 6-minute walk test (6MWT), and physical activity (PA), were measured at baseline and after six weeks. Results: When compared according to cachexia status at baseline, patients with no/pre-cachexia had a mean loss of muscle mass (−5.3 cm2, p = 0.020) but no statistically significant change in PF outcomes. Patients with cachexia also lost muscle mass but to a lesser extent (−2.8 cm2, p = 0.146), but demonstrated a statistically significant decline in PF; KPS (−3.8 points, p = 0.030), self-PF (−8.8 points, p = 0.027), and HGS (−2.7 kg, p = 0.026). Conclusions: Weight loss history and cachexia status at baseline are of importance if one aims to detect changes in PF outcomes in cancer cachexia trials. To improve the use of co-primary endpoints that include PF in future trials, outcomes that have the potential to detect change relative to weight loss should be investigated further.