Background. In the last decades, a growing number of studies investigated the Technology Based Intervention (TBI) (computer, phone, tablet, virtual reality, robot) for supporting children and teenagers with ASD in their daily life, and notably in school settings. Past systematic reviews indicated that these studies are methodologically too weak for any conclusive claim: TBIs are at best promising practices for ASD interventions (e.g., Grynszpan, et al., 2014; Odom, et al., 2015). Objectives. Our purpose was threefold: 1) to update the previous systematic reviews of TBI studies for ASD with a focus on clinical-quality studies; 2) to examine reliability, consistency, durability and generalization of study measures; and 3) to compare the methodology of two cores of studies according to two dimensions: Therapeutic Effectiveness (TE) and Technology Usability (TU). Methods. Following the PICO method, a systematic literature search was conducted using 9 online databases from 2000 to 2016 with keywords related to autism, TE and TU. Resulting studies were selected using well-established quality criteria (SIGN, JADAD). Studies were then classified according to the following methodological dimensions: reliability (standardized- vs. non-standardized scales, objective vs. subjective measures), consistency (internal vs. external measurement of TBI’s target), durability (near/far effect) and generalization (transfer or ecological value). From this stage, TE and TU studies were compared, and those addressing both dimensions were examined separately. Results. From the 917 search results, 31 studies were selected, including 22 on TE, 6 on TU and 3 addressing both dimensions (TE-TU). Overall, TE studies were found more rigorous than TU studies with respect to study design, sample size and ASD-related inclusion/exclusion criteria. TE studies are the only ones with RCT design, and have made more precise description of their samples. Regarding reliability of TBI effect, TE studies employed more standardized measures than TU studies. Objective measures were used in most studies. Regarding consistency of TBI effect, internal validity was respected in both TE and TU studies, whereas external validity was not. Generalization measures were only included in a few TE studies, and absent in TU studies. Durability of TBI effect was mostly occulted, since only two TE studies examined post-intervention, near/far effects. Studies' results were classified as low, mild or high positive, and interestingly, the more the study design was robust, the more the consistency dropped. Finally, the three TE-TU studies were less rigorous than most TE studies, but used standardized or objective measures. Furthermore, they addressed internal validity, but not external validity, durability and generalization. Conclusions. Regardless of the dimension (TE, TU), very few TBI studies for ASD reached the standards of evidence-based practices (reliability, consistency, durability, generalization). Nevertheless, TE studies provided more evidence of their reliability by using more standardized measures, particularly for TBI including computer software, which address emotional and social skills. Studies addressing both TU and TE are an emerging research approach. This situation is promising because TBI usage is a prerequisite to TBI effects. In other words, the clinical benefits of a TBI necessitate usability. This situation calls for interdisciplinary development efforts of TBI for ASD, combining expertise in human-computer interaction and clinical research.