The mechanical performance of expanded polystyrene (EPS) thermal insulation produced with mechanically recycled content has received limited investigation beyond the single point compressive stress at 10% strain (${\mathcal{\sigma }}_{10}$) declared under EN 13163. This study quantifies the cumulative volumetric energy absorption W(ε) and the specific energy absorption SEA(ε) of EPS containing recycled material from two distinct end-of-life sources. Construction waste (S1) and packaging waste (S2) were tested to find whether energy-based metrics reveal source effects masked by σ₁₀. Two nominal density classes, corresponding to the EPS 100 and EPS 150 types, five recycled content levels (0%, 5%, 10%, 25%, 50% by volume), and 235 specimens were tested under EN ISO 29469:2022. A two-way ANOVA framework decomposed the recycling effect into source, content, and interaction components, with Welch’s ANOVA, classical ANOVA, and Kruskal–Wallis applied in parallel as robustness checks. Recycled content was the dominant factor across all metrics. The two types behaved differently with respect to source: EPS 100 showed an additive source effect throughout the dose range, while EPS 150 was dominated by a strong source × content interaction, with the sources equivalent at low doses and S1 substantially outperforming S2 at 50% substitution. A recycled fraction of 5% was statistically equivalent to virgin material in every configuration tested. The findings show that energy-based metrics resolve recycled material source effects that are not detectable on σ₁₀ alone, and that the substitution is type-specific rather than universal.