The impact of intermittent preventive treatment (IPT) on malaria in pregnancy is well known. In countries where this policy is implemented, poor access and low compliance to this intervention has been widely reported. A study was designed to assess a new approach to deliver IPT to pregnant women through traditional birth attendants (TBAs), drug-shop vendors (DSVs), community reproductive health workers (CRHWs) and adolescent peer mobilisers (APMs); and compared this approach with IPT at health units. We evaluated this approach to assess user perceptions, its acceptability and sustainability. Results show that the new approach increased access and compliance to IPT. Mean gestational age at first dose of IPT was 21.0 weeks with the community approaches versus 23.1 weeks at health units, P>0.0001. Health units accessed a high proportion of adolescents, 28.4%, versus 25.0% at the new approaches, P<0.03; most primigravidae, 23.6%, versus 20.0% at the new approaches, P<0.04. The proportion of women who received two doses of SP was 67.5.2% with the new approaches versus 39.9% at health units, P<0.0001. The new approach was associated with a three-fold increase in use of ITNs from 8.8% at baseline to 23.4%. The factors that most influenced acceptability and use of IPT were trusted and easy accessible resource persons, their ability to make home visits especially with CRHWs and APMs; the support of spouses. Another factor was the high awareness on dangers of malaria in pregnancy and the benefits of IPT created by the resource persons. The women perceived better health using the first dose of sulphadoxine-pyremethamine (SP) and this compelled them to go for the second dose. IPT with this approach was highly acceptable with 89.1% of women at the new approaches intending to use it for the next pregnancy, while 48.0% of them had recommended it to other women. We suggest a review of the current policy on malaria prevention in pregnancy to allow provision of IPT through community structures that are feasible, practical and acceptable.