Background: Emerging evidence supports the efficacy of mobile phone interventions for smoking cessation. However, behaviour changes of smokers who fail to reach abstinence and the related psychological mechanism are still understudied. We aimed to evaluate the efficacy of a behaviour change theory-based smoking cessation intervention delivered through personalised text messages from the perspective of smokers who fail to reach abstinence. Methods: We conducted a two-arm, double-blind, randomised controlled trial, with the intervention group receiving personalised text messages developed specifically for this study, and the control group receiving non-personalised ones related to smoking cessation. These messages were sent over a period of three months. We looked at three outcomes: changes in smoking frequency, in smoking amount, and in protection motivation theory (PMT) construct scores. Results: We obtained smoking cessation results for 722 participants who went through the randomisation process (intervention: 360, control: 362; block randomisation design). Overall, 32.3% of baseline daily smokers in the intervention group and 20.0% in the control group changed to weekly smokers during the follow-up period (P < 0.001), while 43.4% of consistent daily smokers in the intervention group and 32.8% in the control group continuously reduced their smoking amount (P < 0.001). We observed associations between the intrinsic rewards of smoking and changes from daily to weekly smoking, the perceived severity of smoking and reductions in smoking amount, as well as the self-efficacy of quitting and changes from daily to weekly smoking/reductions in smoking amount. Conclusions: We found that a personalised text message-based intervention was more likely to decrease the frequency and amount of smoking in smokers who failed to reach abstinence compared to a non-personalised one. We also explored the possible psychological mechanism of such positive effects. Here we provide evidence for countries to consider the promotion of smoking cessations using behaviour theory-driven personalised text messages, which may be more cost-effective than traditional approaches. Registration: Peking University: ChiCTR2100041942.