Due to heterogeneous nonlinear forcing of complex geomorphological features, predictability of monsoon rainfall 10–90-day intra-seasonal oscillations (ISO) over the complex terrain of northeastern and western Himalayan region (NEH and WH) remained poorly quantified. Using 72 and 61 number of station observations of monsoon rainfall ISOs of NEH and WH, respectively, this study attempts to investigate variation in the regional scale predictability of monsoon rainfall ISOs with respect to changing geomorphological features and monsoon rainfall characteristics. In view of the bimodal nonlinear dynamical structure of monsoon rainfall ISO, the fractal dynamical Hurst exponent-based predictability indices are estimated as an indicator of predictability for station observations of NEH and WH, and relationships with elevations, slopes, aspects, and average numbers of occurrences of long (short) spell of active (break) phases are investigated. Results show 10–90-day ISOs are anti-persistent throughout the IHR, although, predictability of 10–90-day ISOs is higher over the NEH region than WH. Predictabilities of ISOs are found to decrease with increasing elevation and slope for both NEH and WH regions. Predictabilities of ISOs over both regions are also found to increase linearly as the number of occurrences of monsoon rainfall ISO phases (active/break) increases.