This systematic analysis examines effectiveness research on online and blended learning from schools, particularly relevant during the Covid-19 pandemic, and also educational games, computer-supported cooperative learning (CSCL) and computer-assisted instruction (CAI), largely used in schools but with potential for outside school. Eight research databases were searched. Studies which were non-school, before 2000, not in English, without data and duplicates were removed, leaving 1355 studies: online 7%, blended 13%, CSCL 7%, games 26% and CAI 47%. Overall, digital technology was more effective (better) than regular instruction in 85% of studies, 8% the same and 3% worse. Blended learning was considerably better than online learning. CAI was the most effective, with games and CSCL coming after blended learning, but of course CAI was not searched for and these were not widely used outside of schools. Primary and early years/kindergarten were most effective (87% better) and secondary/high next (80%). Although science and mathematics were the most popular subjects, English as a foreign language interventions were the most effective, then writing and STEM, thinking, arts/music, humanities, health and science, reading and maths, foreign languages and English in that order. Overall, females did better than males. ‘Low ability’ children and second language learners did especially well. Disadvantaged and special educational needs/disabled students did slightly worse. Self-efficacy was highly related to better outcomes. The limitations/strengths of the research were discussed and linked back to previous literature, a critical analysis made, and detailed implications for practitioners, policy makers and future researchers outlined. Digital technology's main advantage may be the possibility for enhanced task flexibility and learner autonomy, encouraging greater self-regulation. However, this may not be an advantage for all students.