Half of the tributaries of the Amazon River originate in the tropical Andes; however, it is difficult to assess hydroclimatic conditions due to the scarcity of long, high-quality instrumental records. Data from the Global Precipitation Climatology Project (GPCP) provides a complete record since 1979 and offers a good representation of rainfall over the tropical Andes. Longer records are needed to improve our understanding of rainfall variability and summer monsoon behavior at various scales. We developed the first annually-resolved precipitation reconstruction for the tropical Andes in Peru, based on tree-ring chronologies of Cedrela and Juglans species. The annual (November-October) reconstruction extends the short instrumental records back to 1817, explaining 68% of the total variance of precipitation over the 1979-2007 calibration period. The reconstruction reveals the well-documented influence of ENSO on Amazon rainfall at interannual scales (~19% of total variance), and significant multidecadal variability with alternating periods of about 40 years (~13% of rainfall variability) related to the Atlantic Multidecadal Oscillation (AMO). Both oscillatory modes can explain dry and humid periods observed within the reconstruction and are likely associated with the negative trends of rainfall in the short instrumental records and the increased drought recurrence in recent decades. Our results show that montane tropical tree rings can be used to reconstruct precipitation with exceptionally high fidelity, characterize the interannual to multidecadal variability and identify remote forcings in the hydroclimate over the Andean Amazon basin of Peru.