The impact of dual tasks on gait in Parkinson's disease (PD) reveals lack of automaticity and increased cognitive demands. We explored which characteristics explained walking speed with and without dual task interference and if they reflected the cognitive demands of the task. In 130 people with PD, gait performance was quantified in the home using accelerometers allowing estimates of single and dual task walking speed and interference (difference between dual and single task). Multiple regression analysis was used to explore the effect of 12 characteristics representing four domains (personal, motor symptoms, cognitive, affective) on gait outcomes. Thirty-seven percent of variance in single task speed was explained by increased fear of falling, sex, age, disease severity, and depression; 34% of variance in dual task speed was explained by increased fear of falling, disease severity, medication, and depression; 12% of variance in interference scores was explained by greater disease severity and impaired executive function. Personal, motor, affective, and cognitive characteristics contribute to walking speed and interference, highlighting the multifactorial nature of gait. Different patterns of characteristics for each outcome indicates the impact of cognitive demand and task complexity, providing cautious support for dual task speed and interference as valid proxy measures of cognitive demand in PD gait.