This study investigates the effectiveness of donation matching gift schemes using lotteries for increasing charitable giving relative to deterministic matches of equivalent or higher expected value and a no-match control. We recruit 1,402 online participants and randomly assign them to one of seven conditions: No Match (Control), and two sets of matching schemes of varying equivalent expected values: EV=1 (1:1 Deterministic Match, 1:10% of 10 tokens, and 1:1% of 100) and EV=0.5 (2:1 Deterministic Match, 1:1% of 50 tokens and 1:0.5% of 100), where one token is $0.50. Participants complete three 10-token allocation decisions for hunger-related charities with one allocation randomly selected for realization. The 1:1 Match significantly increases giving by 15.7% compared to No Match. We find that matching schemes with a small probability of a very large amount (1% and 0.5% of 100) elicit significantly higher rates of giving compared to No Match (Mann-Whitney p=0.019 and p=0.096 respectively) and do not statistically differ from the 1:1 Match (Mann-Whitney p=0.976 and p=0.622 respectively). Our results suggest nonprofits can use matching gift donations more efficiently through lottery matching donation schemes while increasing downstream donations.