Background: Non-cardiac chest pain is common with two-thirds due to gastroesophageal reflux disease (GERD). Objective: To evaluate the effectiveness and safety of guided vs. empirical therapy in non-cardiac chest pain. Methods: Adults with normal angiogram or stress test were randomized into either a guided or empirical group. In the guided group, after the ambulatory pH-impedance test, if GERD then dexlansoprazole 30 mg/day for 8 weeks, but if functional or hypersensitive chest pain, then theophylline SR 250 mg/day for 4 weeks. In the empirical group, dexlansoprazole 60 mg/day was given for 2 weeks. The primary outcome was global chest pain visual analog score (VAS) and secondary outcomes were Quality of Life in Reflux and Dyspepsia (QOLRAD), GERD questionnaire (GERDQ), and pH parameters, all determined at baseline, 2nd and 8th weeks. Results: Of 200 screened patients, 132 were excluded, and of 68 randomized per-protocol, 33 were in the guided group and 35 in the empirical group. For between-group analysis, mean global pain scores were better with guided vs. empirical group at 8th week (P = 0.005) but not GERDQ or QOLRAD or any of pH measures (all P > 0.05). For within-group analysis, mean QOLRAD improved earliest at 8th week vs. baseline (P = 0.006) in the guided group and 2nd week vs. baseline (P = 0.011) in the empirical group but no differences were seen in other secondary outcomes (P > 0.05). No serious adverse events were reported. Conclusions: Guided approach may be preferred over short-term empirical therapy in symptom response, however QOLRAD, acid-related symptoms, or pH measures are not significantly different (trial registration ID no. NCT03319121).