The Tektronix TDS-2004b units feature a vertical sensitivity of 2mV/Div. This is with 8 divisions so a total screen of 16mV. Note however that the bandwidth of the unit goes down to 20 MHz when you choose this range.
The majority of analog and digital scopes have a lowest sensitivity level of around 1mV-50mV/div. Some are significantly higher such as 50V/div. 10uV/div, while not impossible, would be very rare.
While voltage ranges are a important factor to figure in when researching a new oscilloscope, resolution, buffer size, bandwidth, sampling rate, ease of use etc are all also equally important.
It depends on your application which unit would be the best for you. For example, my applications typically do not involve measuring voltage as low as 10uV's, so I don't need to have a range that low. If I did need it, I would probably find a way to amplify the signal.
Crag Technologies, Inchttp://www.pc-oscilloscopes.com
Problem with amplification of the signal (lets say 10uV becomes 1mV, or a gain of 100), is that the scope will report 1mV, even though the signal I'm trying to measure is only 10uV. So my readout is inacurate. But then the scope software would have to compensate. Even more, such signal pre-amp wouldn't have a NIST calibration if I made one myself, so 10uV might ACTUALLY get me 1.1mV or 0.9mV, instead of the desired 1mV. And it might drift over time from giving 1.0V to 1.1V because I (as an individule rather than a company) wouldn't have the best parts. Lastly, the simplest amplifier to make is with an op-amp chip. But due to finite slew rate these are only linear up to about 20kHz and that ruins specs of my oscilloscope then. You mentioned the TDS-2004b does get better resolutions of less uV/pixel than Picoscope, and when in this mode limits its frequency response. Do you think you could implement that in you Picoscopes?