This research was conducted in 2022 to describe a picture of poverty, feed behavior, and the incidence of stunting in the city of Semarang. Method: The data collection method uses structured interviews with 384 samples of families at risk of stunting using the proportional random sampling method. Methods of processing and analyzing data with descriptive statistics (Chi-Square). Results: The results showed that the number of respondents in the poor category was 238 respondents (62.0%) consisting of 224 respondents (58.3%) who had stunted toddlers and as many as 14 respondents (3.6%) who did not have stunted toddlers. Meanwhile, 146 respondents (38.0%) were in the non-poor category, consisting of 115 respondents (29.9%) who had stunted children under five and 31 respondents (8.1%) who did not have stunted children. The results of data analysis using the Chi-Square test obtained a p-value of 0.000 (p > 0.05), which means that there is a relationship between poverty and stunting. Conclusion: Respondents in the poor category are 0.232 times more likely to have stunting toddlers. In addition, there was also a difference in the feeding behavior of toddlers in families at risk of stunting in the city of Semarang