The claim that insecure attachment predisposes individuals to anxiety was put forth by attachment theorists more than 45 years ago, yet evidence of this association has produced mixed results. By conducting a series of meta-analyses (k = 53, N = 4,970), we found that individuals with secure and insecure attachment representations did not differ in their reported anxiety symptoms. However, individuals with preoccupied (d = 0.35) but not dismissing attachment representations (d = −0.02) endorsed significantly more anxiety symptoms than secure individuals. Individuals with preoccupied attachment representations also endorsed greater anxiety symptoms than people with dismissing attachment representations (d = 0.31), and this difference was stronger when assessed in high-risk samples than low-risk samples. Unresolved individuals were more anxious than nonunresolved individuals (d = 0.29), but exploratory analysis suggested that secondary preoccupied attachment subtype drove this effect. Results highlight the relevance of attachment representations for prevention and intervention efforts targeting anxiety.