We investigated comparison according to reflux esophagitis and non-erosive reflux disease about "daily" symptom improvement for proton pump inhibitor treatment. We enrolled 57 reflux esophagitis and 90 non-erosive reflux disease patients. They took rabeprazole 10 mg/day for 28 days and completed "daily" in the Frequency Scale for the Symptoms of GERD from baseline until day 14, and after 28 days of treatment. The efficacy endpoint was the improvement rates in Frequency Scale for the Symptoms of GERD, based on baseline. Frequency Scale for the Symptoms of GERD was decreased in reflux esophagitis and non-erosive reflux disease (p<0.001) and was significantly lower in reflux esophagitis than in non-erosive reflux disease from the first day of treatment (p<0.05). Symptomatic improvement rates were also significantly higher in reflux esophagitis (50.3 ± 44.9%) than in non-erosive reflux disease (31.7 ± 43.2%) from the first day of treatment (p<0.0001). The symptomatic improvement rates in reflux esophagitis were significant increased from the second day of treatment until after 28 days of treatment (p = 0.0006), however, these in non-erosive reflux disease were significant increased from third days until after 28 days of treatment (p = 0.0002). In non-erosive reflux disease, the improvement of dysmotility symptom was particularly gradual as well as of reflux symptom, too. As for results of prediction of proton pump inhibitor response (completed symptom resolution) form early symptom improvement within 1 week, it was able to predict proton pump inhibitor response from the symptom improvement rate on 3 days in reflux esophagitis and on day 7 in non-erosive reflux disease. In conclusion, the prediction of the proton pump inhibitor response in non-erosive reflux disease was slow in comparison with reflux esophagitis. The cause was gradual improvement of dysmotility symptom.