This field study investigated the phytoremediation potential of two arsenic (As) hyperaccumulating fern species, Pityrogramma calomelanos var. austroamericana and Pteris vittata over 27-month duration at a disused As-contaminated cattle-dip site located at Wollongbar, NSW, Australia. Ferns planted in January 2009 were harvested following 10, 22 and 27 months of growth. A detailed soil sampling was undertaken in June 2009 (initial, n = 42 per plot) and limited sampling in April 2011 (after 27 months, n = 15 per plot) to measure total and phosphate-extractable As concentrations in soil at 0 − 20-, 20 − 40- and 40 − 60-cm depths. The choice of the limited number of samples was considered sufficient to estimate the changes in soil As concentration following phytoremediation based on a geostatistical model. The average frond dry biomass, As concentration and As uptake were significantly (P 0.05), respectively, by P. vittata. Our results show that phytoremediation time based on observed changes in soil As based on limited sampling is not reliable; hence, it is recommended that the frond As uptake should be considered in order to evaluate the phytoremediation efficiency of the two fern species at the experimental site. Using As uptake of the two fern species, we estimate that with P. calomelanos var. austroamericana it would take 55 − 125 years to decrease mean total As content below the ecological investigation level (20 mg kg−1) in the surface and subsurface soils, whereas with P. vittata 143 − 412 years would be required to achieve this target.