The treatment gap for autism globally is high. Our previous PASS intervention, delivered by community based lay health workers, showed effectiveness. This article reports the development and evaluation of a new "PASS 'Plus'" intervention in a rural population in India. Using formative research methods, we supplemented the PASS intervention with additional (Plus) modules to address autism comorbidities. This is the first time that a rigorous methodology has been used to evaluate autism symptom outcomes in a low and middle-income country setting. 40 parent-child dyads were recruited in a pilot randomized controlled trial against usual care (mean age 65 months (34 boys); n = 19 PASS Plus, n = 21 UC). 89% of intervention families partially or entirely completed the 12-session intervention. Intention to treat analysis showed a reduction in mean scores of autism symptom severity, though the confidence interval contains zero, (adjusted mean difference AMD -2.42 95% CI -7.75, 2.92; ES 0.22); large treatment effects on proximal outcomes of proportion of parent synchronous responses (AMD 0.35; 95% CI 0.18, 0.52; effect size ES 3.97) and proportion of child communication initiations with parent (AMD 0·17; 95% CI 0.03, 0.32; ES 1.02). Confidence intervals for effects on mutual shared attention (AMD 0.10; 95% CI -0.07, 0.27; ES 0.5) and co-morbid symptoms (AMD -9.0; 95% CI -24.26, 6.26; ES 0.32) contained zero. There were significant effects to improve parental mental health. PASS Plus shows good feasibility and adds to the evidence of the effectiveness of task sharing complex autism interventions to lay health workers in India.