To provide a clear landscape, trends, and research frontiers of gene therapy, we systematically retrieved a total of 62,961 peer-viewed studies published between 1996 and 2020 from the Scopus, Web of Science, and 42,120 Inpadoc patent families from Derwent Innovation databases. Multiple bibliometric approaches suggest that gene therapy began to recover in 2013 after a period of significant decline. However, metrics in terms of authors and scholarly output growth, FWCI, annual citations, percentage of high-impact journal literature, and patent-citations per scholarly output are still weak at this stage, indicating a lack of research momentum. We also visualized gene therapy's knowledge structure by employing citation analysis, co-citation analysis, and co-word analysis, revealing its research hotspots and trends by text mining with Natural Language Processing. For the current predicament, we propose that the future success of gene therapy may depend on breakthroughs in more advanced and exhilarating technologies such as the CRISPR-Cas system, CAR-T cell therapies, and gene delivery vector technology. The results show that evidence-based bibliometrics allows the dissection of gene therapy to inform scientific planning and decision-making.