Abstract ICESat‐2 has demonstrated solid performance in retrieving sea ice freeboard during winter, when sea ice surface types are well‐defined and leads provide reliable sea surface height references. Summer retrievals are challenging because melt ponds are not explicitly distinguished in the ATL07 surface classification and may be identified as ice or leads. Here we assess ICESat‐2 summer sea ice observations using airborne LiDAR and imagery collected along near‐coincident ground tracks during a July 2022 calibration and validation campaign. Over a ∼50‐km Arctic sea ice transect, we compare ICESat‐2 surface classifications, elevations, and freeboards with high‐resolution airborne measurements. Melt ponds are generally not selected as sea surface height references in the ICESat‐2 freeboard retrieval. ICESat‐2 and airborne surface heights agree well (R ≥ ${ge} $ 0.9) and there are minimal freeboard differences (−0.05 ± $pm $ 0.12 m to 0.01 ± $pm $ 0.10 m). ICESat‐2 summer freeboard retrievals are largely unaffected by melt pond misclassification, supporting the potential for year‐round sea ice thickness monitoring.