This paper addresses the problem of predicting crack development at a stress concentration feature from simple laboratory crack growth data. It uses a specifically designed double edge notch geometry as a test vehicle and growth rates generated on standard corner crack and compact tension specimens. A coarse-grained, near-alpha titanium alloy is employed to emphasize any microstructural interaction effects. The bulk of the work is at ambient temperature with a small amount of data obtained at 300°C. It is shown that at low stress (350 MPa), conventional fracture mechanics, without an allowance for crack closure, can correctly predict both crack shape development and cyclic life. The only significant errors occur when the crack front is retarded by prior beta or alpha colony boundaries. These tend to lead to pessimistic predictions. At 550 MPa, however, there is significant notch root plasticity that induces crack closure. An allowance for notch closure in the numerical procedures permits a more accurate prediction of crack shape and cyclic life at this stress level. To support the computer models, a detailed assessment of closure is made using potential difference (PD), strain gage, and replica techniques.