Study design: Cross-sectional. Objectives: Individualized training regimes are often based on ventilatory thresholds (VTs). The objectives were to study: (1) whether VTs during arm ergometry could be determined in individuals with spinal cord injury (SCI), (2) the intrarater and interrater reliability of VT determination. Setting: University research laboratory. Methods: Thirty graded arm crank ergometry exercise tests with 1-min increments of recreationally active individuals (tetraplegia (N = 11), paraplegia (N = 19)) were assessed. Two sports physicians assessed all tests blinded, randomly, in two sessions, for VT1 and VT2, resulting in 240 possible VTs. Power output (PO), heart rate (HR), and oxygen uptake (VO2) at each VT were compared between sessions or raters using paired samples t-tests, Wilcoxon signed-rank tests, intraclass correlation coefficients (ICC, relative agreement), and Bland–Altman plots (random error, absolute agreement). Results: Of the 240 VTs, 217 (90%) could be determined. Of the 23 undetermined VTs, 2 (9%) were VT1 and 21 (91%) were VT2; 7 (30%) among individuals with paraplegia, and 16 (70%) among individuals with tetraplegia. For the successfully determined VTs, there were no systematic differences between sessions or raters. Intrarater and interrater ICCs for PO, HR, and VO2 at each VT were high to very high (0.82–1.00). Random error was small to large within raters, and large between raters. Conclusions: For VTs that could be determined, relative agreement was high to very high, absolute agreement varied. For some individuals, often with tetraplegia, VT determination was not possible, thus other methods should be considered to prescribe exercise intensity.