Abstract. The temporal scaling properties of F-region velocity fluctuations, δv los , were characterised over 17 octaves of temporal scale from τ =1 s to <1 day using a new data base of 1-s time resolution SuperDARN radar measurements. After quality control, 2.9 (1.9) million fluctuations were recorded during 31.5 (40.4) days of discretionary mode soundings using the Tasmanian (New Zealand) radars. If the fluctuations were statistically self-similar, the probability density functions (PDFs) of δv los would collapse onto the same PDF using the scaling P s (δv s , τ )=τ α P (δv los , τ ) and δv s =δv los τ −α where α is the scaling exponent. The variations in scaling exponents α and multi-fractal behaviour were estimated using peak scaling and generalised structure function (GSF) analyses, and a new method based upon minimising the differences between re-scaled probability density functions (PDFs). The efficiency of this method enabled calculation of "α spectra", the temporal spectra of scaling exponents from τ =1 s to ∼2048 s. The large number of samples enabled calculation of α spectra for data separated into 2-h bins of MLT as well as two main physical regimes: Population A echoes with Doppler spectral width <75 m s −1 concentrated on closed field lines, and Population B echoes with spectral width >150 m s −1 concentrated on open field lines. For all data there was a scaling break at τ ∼10 s and the similarity of the fluctuations beneath this scale may be related to the large spatial averaging (∼100 km×45 km) employed by SuperDARN radars. For Tasmania Population B, the velocity fluctuations exhibited approximately mono fractal power law scaling between τ ∼8 s and 2048 s (34 min), and probably up to several hours. The scaling exponents were generally less than that expected for basic MHD turbulence (α=0.25), except close to magnetic dusk where they peaked towards the basic MHD value. For Population A, the scaling exponents were larger than for Population B, having values generally Correspondence to: M. L. Parkinson (m.parkinson@latrobe.edu.au) in the range expected for basic MHD and Kolmogorov turbulence (α=0.25-0.33). The α spectra exhibited complicated variations with MLT and τ which must be related to different physical processes exerting more or less influence.