Telehealth services expanded rapidly during the COVID-19 public health emergency (PHE). To evaluate changes in availability of telehealth services at outpatient mental health treatment facilities (MHTFs) throughout the US during and after the COVID-19 PHE. In this cohort study, callers posing as prospective clients contacted a random sample of 1404 MHTFs drawn from the Substance Abuse and Mental Health Services Administration's Behavioral Health Treatment Locator from December 2022 to March 2023 (wave 1 [W1]; during PHE). From September to November 2023 (wave 2 [W2]; after PHE), callers recontacted W1 participants. Analyses were conducted in January 2024. Callers inquired whether MHTFs offered telehealth (yes vs no), and, if yes, whether they offered (1) audio-only telehealth (vs audio and video); (2) telehealth for therapy, medication management, and/or diagnostic services; and (3) telehealth for comorbid alcohol use disorder (AUD). Sustainers (offered telehealth in both waves), late adopters (did not offer telehealth in W1 but did in W2), nonadopters (did not offer telehealth in W1 or W2), and discontinuers (offered telehealth in W1 but not W2) were all compared. During W2, 1001 MHTFs (86.1%) were successfully recontacted. A total of 713 (71.2%) were located in a metropolitan county, 151 (15.1%) were publicly operated, and 935 (93.4%) accepted Medicaid as payment. The percentage offering telehealth declined from 799 (81.6%) to 765 (79.0%) (odds ratio [OR], 0.84; 95% CI, 0.72-1.00; P < .05). Among MHTFs offering telehealth, a smaller percentage in W2 offered audio-only telehealth (369 [49.3%] vs 244 [34.1%]; OR, 0.53; 95% CI, 0.44-0.64; P < .001) and telehealth for comorbid AUD (559 [76.3%] vs 457 [66.5%]; OR, 0.62; 95% CI, 0.50-0.76; P < .001) compared with W1. In W2, MHTFs were more likely to report telehealth was only available under certain conditions for therapy (141 facilities [18.0%] vs 276 [36.4%]; OR, 2.62; 95% CI, 1.10-3.26; P < .001) and medication management (216 facilities [28.0%] vs 304 [41.3%]; OR, 1.81; 95% CI, 1.48-2.21; P < .001). A total of 684 MHTFs (72.0%) constituted sustainers, 94 (9.9%) were discontinuers, 106 (11.2%) were nonadopters, and 66 (7.0%) were late adopters. Compared with sustainers, discontinuers were less likely to be private for-profit (adjusted OR [aOR], 0.28; 95% CI, 0.11-0.68) or private not-for-profit (aOR, 0.26; 95% CI, 0.14-0.48) after adjustment for facility and area characteristics. Based on this longitudinal cohort study of 1001 MHTFs, telehealth availability has declined since the PHE end with respect to scope and modality of services, suggesting targeted policies may be necessary to sustain telehealth access.