Over the last five decades a growing number of governments in developed and developing countries have implemented targeted policies to increase the R&D to GDP ratio. However, there is little evidence regarding the feasibility of achieving large and permanent changes in R&D investment. We study the incidence and effects of episodes of substantial acceleration in R&D expenditure, using a sample of 62 countries with data from 1960 to 2007. Among other exercises, we use propensity score matching, synthetic cohorts, and panel VAR, in order to elucidate the determinants and effects of important increases in R&D intensity. We find that transitions to higher levels of R&D-intensity are a relatively infrequent phenomenon which occurs at relatively high levels of R&D intensity. Looking at long-run changes in R&D, we corroborate that few countries have been able to raise their R&D intensity from the bottom quintile to the top quintile of the global distribution. Our findings indicate that income, physical investment, education, and the size of the manufacturing sector increase the likelihood of transition, whereas country size and FDI decrease it. We find that transitions are positively correlated with subsequent greater income levels, and weakly to TFP growth. Finally, in our Granger tests with panel VAR estimates, we find that R&D acceleration Granger causes GDP growth, the level of patents, high-tech exports, and private and public R&D. In the case of private R&D, there is evidence of bidirectional causality with R&D acceleration.